feat: add repo-memory to all agentic workflows for cross-run editorial context#939
feat: add repo-memory to all agentic workflows for cross-run editorial context#939
Conversation
…rial context Instead of adding generic MCP servers (server-memory, sequential-thinking), this uses gh-aw's built-in repo-memory tool which is superior: - Persistent across runs (git-backed, unlimited retention) - Shared branch allows all workflows to read each other's context - Tracks generated articles to avoid duplicates - Maintains editorial continuity across runs Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/b711fddd-f5b1-47ae-8e25-c9ebf97f19be Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
There was a problem hiding this comment.
Pull request overview
Adds gh-aw repo-memory support to the news-generation agentic workflow sources to preserve editorial/translation context across workflow runs (via a shared memory/news-generation branch).
Changes:
- Added
tools.repo-memoryconfiguration to all 10news-*.mdgh-aw workflows (shared branchmemory/news-generationwith.md/.jsonallowlist and size limits). - Added a “Repo Memory — Cross-Run Editorial Context” instruction section to each workflow prompt describing read-at-start and write-at-end patterns.
Reviewed changes
Copilot reviewed 10 out of 10 changed files in this pull request and generated 20 comments.
Show a summary per file
| File | Description |
|---|---|
| .github/workflows/news-breaking.md | Adds repo-memory tool config and breaking-news memory read/write instructions. |
| .github/workflows/news-weekly-review.md | Adds repo-memory tool config and weekly editorial continuity instructions. |
| .github/workflows/news-week-ahead.md | Adds repo-memory tool config and cross-run planning context instructions. |
| .github/workflows/news-monthly-review.md | Adds repo-memory tool config and monthly editorial continuity instructions. |
| .github/workflows/news-month-ahead.md | Adds repo-memory tool config and cross-run strategic outlook context instructions. |
| .github/workflows/news-committee-reports.md | Adds repo-memory tool config and cross-run committee coverage context instructions. |
| .github/workflows/news-propositions.md | Adds repo-memory tool config and cross-run legislative procedure coverage context instructions. |
| .github/workflows/news-motions.md | Adds repo-memory tool config and cross-run plenary/votes coverage context instructions. |
| .github/workflows/news-article-generator.md | Adds repo-memory tool config and shared editorial context instructions for manual generation workflow. |
| .github/workflows/news-translate.md | Adds repo-memory tool config and translation-progress log instructions. |
| tools: | ||
| repo-memory: | ||
| branch-name: memory/news-generation | ||
| description: "Cross-run editorial memory for EU Parliament news generation" | ||
| file-glob: ["memory/news-generation/*.md", "memory/news-generation/*.json"] |
There was a problem hiding this comment.
These frontmatter changes won’t affect the actual GitHub Actions run until the corresponding .lock.yml is recompiled and committed (the repo executes the compiled lock workflow). Consider recompiling the gh-aw lock files as part of this PR (or triggering the compile workflow) so repo-memory is actually enabled on merge.
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | ||
| ``` | ||
|
|
||
| **At workflow END** — update memory (keep concise, max 50KB total): |
There was a problem hiding this comment.
The text “max 50KB total” is misleading given the repo-memory config limits are per-file/per-patch (and allow multiple files). Suggest rewording to clarify the limit being per file (or align the tool limits if you truly want a total cap).
| **At workflow END** — update memory (keep concise, max 50KB total): | |
| **At workflow END** — update memory (keep concise, max 50KB per file): |
| tools: | ||
| repo-memory: | ||
| branch-name: memory/news-generation | ||
| description: "Cross-run editorial memory for EU Parliament news generation" | ||
| file-glob: ["memory/news-generation/*.md", "memory/news-generation/*.json"] |
There was a problem hiding this comment.
These frontmatter changes won’t affect the actual GitHub Actions run until the corresponding .lock.yml is recompiled and committed (the repo executes the compiled lock workflow). Consider recompiling the gh-aw lock files as part of this PR (or triggering the compile workflow) so repo-memory is actually enabled on merge.
.github/workflows/news-week-ahead.md
Outdated
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | ||
| ``` | ||
|
|
||
| **At workflow END** — update memory (keep concise, max 50KB total): |
There was a problem hiding this comment.
The text “max 50KB total” is misleading given the repo-memory config limits are per-file/per-patch (and allow multiple files). Suggest rewording to clarify the limit being per file (or align the tool limits if you truly want a total cap).
| **At workflow END** — update memory (keep concise, max 50KB total): | |
| **At workflow END** — update memory (keep concise; max 50KB per file): |
| tools: | ||
| repo-memory: | ||
| branch-name: memory/news-generation | ||
| description: "Cross-run editorial memory for EU Parliament news generation" | ||
| file-glob: ["memory/news-generation/*.md", "memory/news-generation/*.json"] |
There was a problem hiding this comment.
These frontmatter changes won’t affect the actual GitHub Actions run until the corresponding .lock.yml is recompiled and committed (the repo executes the compiled lock workflow). Consider recompiling the gh-aw lock files as part of this PR (or triggering the compile workflow) so repo-memory is actually enabled on merge.
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | ||
| ``` | ||
|
|
||
| **At workflow END** — update memory (keep concise, max 50KB total): |
There was a problem hiding this comment.
The text “max 50KB total” is misleading given the repo-memory config limits are per-file/per-patch (and allow multiple files). Suggest rewording to clarify the limit being per file (or align the tool limits if you truly want a total cap).
| **At workflow END** — update memory (keep concise, max 50KB total): | |
| **At workflow END** — update memory (keep concise; aim to keep each memory file under 50KB): |
| tools: | ||
| repo-memory: | ||
| branch-name: memory/news-generation | ||
| description: "Cross-run editorial memory for EU Parliament news generation" | ||
| file-glob: ["memory/news-generation/*.md", "memory/news-generation/*.json"] |
There was a problem hiding this comment.
These frontmatter changes won’t affect the actual GitHub Actions run until the corresponding .lock.yml is recompiled and committed (the repo executes the compiled lock workflow). Consider recompiling the gh-aw lock files as part of this PR (or triggering the compile workflow) so repo-memory is actually enabled on merge.
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | ||
| ``` | ||
|
|
||
| **At workflow END** — update memory (keep concise, max 50KB total): |
There was a problem hiding this comment.
The text “max 50KB total” is misleading given the repo-memory config limits are per-file/per-patch (and allow multiple files). Suggest rewording to clarify the limit being per file (or align the tool limits if you truly want a total cap).
| **At workflow END** — update memory (keep concise, max 50KB total): | |
| **At workflow END** — update memory (keep concise, max 50KB per file): |
| tools: | ||
| repo-memory: | ||
| branch-name: memory/news-generation | ||
| description: "Cross-run editorial memory for EU Parliament news generation" | ||
| file-glob: ["memory/news-generation/*.md", "memory/news-generation/*.json"] |
There was a problem hiding this comment.
These frontmatter changes won’t affect the actual GitHub Actions run until the corresponding .lock.yml is recompiled and committed (the repo executes the compiled lock workflow). Consider recompiling the gh-aw lock files as part of this PR (or triggering the compile workflow) so repo-memory is actually enabled on merge.
.github/workflows/news-translate.md
Outdated
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/translation-log.json 2>/dev/null || echo '[]' | ||
| ``` | ||
|
|
||
| **At workflow END** — update memory (keep concise, max 50KB total): |
There was a problem hiding this comment.
The text “max 50KB total” is misleading given the repo-memory config limits are per-file/per-patch (and allow multiple files). Suggest rewording to clarify the limit being per file (or align the tool limits if you truly want a total cap).
| **At workflow END** — update memory (keep concise, max 50KB total): | |
| **At workflow END** — update memory (keep concise, max 50KB per memory file): |
…c workflows Both servers were installed in copilot-setup-steps.yml and configured in copilot-mcp.json but never declared in workflow frontmatter, making them unavailable at runtime in compiled .lock.yml files. Changes: - Added memory + sequential-thinking to mcp-servers: in all 10 news-*.md files - Enhanced prompt sections with usage guidance for all three memory/reasoning tools: - Repo Memory (persistent, cross-run, git-backed) - Memory MCP (session-scoped knowledge graph for batch processing) - Sequential Thinking (structured reasoning chains for complex analysis) Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/ac0d2494-d445-40da-bda2-3d44152d0902 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
…per gh-aw best practices - Remove overly broad *.com, *.org, *.io wildcards from 8 workflow network allowlists (violates AWF firewall best practice of "avoid wildcards") - Add title-prefix "[news] " to create-pull-request safe-output in all 10 workflows - Add max: 1 to add-comment safe-output in all 10 workflows - Update github-agentic-workflows.md skill with current frontmatter patterns, network security best practices, and enhanced safe-output constraint docs - Update gh-aw-firewall.md skill with ecosystem identifier guidance Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/1d534c35-cd01-43f8-84b5-e1f76700a0d8 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
…I analysis guides - Add hack23.com, www.hack23.com, riksdagsmonitor.com, www.riksdagsmonitor.com to network allowlists and safe-outputs allowed-domains in all 10 workflows - Enhance news-translate.md with per-language EP terminology standards table, translation quality dimensions (5-point scoring), and detailed language-specific requirements for Nordic, EU Core, RTL, and CJK language groups - Add multi-language quality standards section to ai-driven-analysis-guide.md with translation fidelity rules, EP terminology table, and Hack23 ecosystem cross-reference documentation - Enhance political-style-guide.md with translation-friendly writing rules (short sentences, explicit subjects, cultural neutrality) and translation quality checklist for all 13 target languages - Update gh-aw-firewall.md and github-agentic-workflows.md skills with Hack23 ecosystem domain documentation and examples Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/9b564e9b-5cb3-4601-a179-d77f8ec3faa5 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/9b564e9b-5cb3-4601-a179-d77f8ec3faa5 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
…uts across all workflows and docs Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/3b48463e-a5e2-4d6e-a334-c6d7df6668c9 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
| network: | ||
| allowed: | ||
| - node | ||
| - github.com | ||
| - api.github.com |
There was a problem hiding this comment.
These frontmatter changes (tightened network allowlist, added repo-memory + MCP servers, updated safe-outputs constraints) won’t take effect until the corresponding .lock.yml is regenerated and committed. Right now the tracked news-weekly-review.lock.yml still contains broad *.com/*.org/*.io allowlists, so merging this without recompiling leaves the executed workflow with the old (less secure) firewall config.
| network: | ||
| allowed: | ||
| - node | ||
| - github.com | ||
| - api.github.com |
There was a problem hiding this comment.
These frontmatter changes won’t affect the workflow that actually runs in GitHub Actions until news-week-ahead.lock.yml is regenerated (via gh aw compile) and committed. The current lock file still carries the old broad *.com/*.org/*.io allowlists, so the firewall tightening in this PR isn’t applied yet.
| @@ -53,9 +53,12 @@ network: | |||
| - api.github.com | |||
| - data.europarl.europa.eu | |||
| - "*.europa.eu" | |||
There was a problem hiding this comment.
This workflow’s .md frontmatter updates (repo-memory, MCP servers, allowlists, safe-outputs constraints) require a recompile to update news-translate.lock.yml. As-is, the lock file still includes the previous broad *.com/*.org/*.io firewall entries, so the executed workflow remains over-permissive.
| - "*.europa.eu" |
| @@ -35,9 +35,12 @@ network: | |||
| - data.europarl.europa.eu | |||
| - api.worldbank.org | |||
| - "*.europa.eu" | |||
There was a problem hiding this comment.
Please regenerate and commit news-propositions.lock.yml as part of this PR. Without recompilation, GitHub Actions will keep running the existing lock workflow which still contains the old broad *.com/*.org/*.io allowlist entries (and won’t include repo-memory / new MCP servers).
| - "*.europa.eu" | |
| - europarl.europa.eu | |
| - www.europarl.europa.eu |
| @@ -34,9 +34,12 @@ network: | |||
| - api.github.com | |||
| - data.europarl.europa.eu | |||
| - "*.europa.eu" | |||
There was a problem hiding this comment.
These security/behavior changes in the markdown source won’t apply to the running workflow until news-motions.lock.yml is regenerated and committed. The current lock file still includes broad *.com/*.org/*.io allowlists, so merging this without recompiling leaves the executed workflow configuration unchanged.
| - "*.europa.eu" |
| @@ -35,9 +35,12 @@ network: | |||
| - data.europarl.europa.eu | |||
| - api.worldbank.org | |||
| - "*.europa.eu" | |||
There was a problem hiding this comment.
Please recompile and commit news-monthly-review.lock.yml so the tightened allowlist, repo-memory, and safe-outputs constraints are actually used by GitHub Actions. The existing lock file still contains broad *.com/*.org/*.io allowlists, so the firewall hardening in this PR isn’t applied yet.
| - "*.europa.eu" |
| @@ -35,9 +35,12 @@ network: | |||
| - data.europarl.europa.eu | |||
| - api.worldbank.org | |||
| - "*.europa.eu" | |||
There was a problem hiding this comment.
This .md update needs a gh aw compile (and committing the result) to update news-month-ahead.lock.yml. Until then, the executed lock workflow still contains broad *.com/*.org/*.io allowlists and won’t include the new repo-memory / MCP server configuration.
| - "*.europa.eu" |
| @@ -34,9 +34,12 @@ network: | |||
| - api.github.com | |||
| - data.europarl.europa.eu | |||
| - "*.europa.eu" | |||
There was a problem hiding this comment.
Please regenerate and commit news-committee-reports.lock.yml as part of this PR. Otherwise GitHub Actions will keep running the existing lock workflow which still includes broad *.com/*.org/*.io allowlists (and will not pick up repo-memory / updated safe-outputs constraints).
| - "*.europa.eu" |
| - hack23.com | ||
| - www.hack23.com | ||
| - riksdagsmonitor.com | ||
| - www.riksdagsmonitor.com | ||
| - euparliamentmonitor.com | ||
| - www.euparliamentmonitor.com |
There was a problem hiding this comment.
This workflow’s .md frontmatter changes require updating news-article-generator.lock.yml via gh aw compile. The current lock file still contains broad *.com/*.org/*.io allowlists, so merging this without committing the regenerated lock file leaves the executed workflow over-permissive and missing repo-memory/sequential-thinking.
| - hack23.com | |
| - www.hack23.com | |
| - riksdagsmonitor.com | |
| - www.riksdagsmonitor.com | |
| - euparliamentmonitor.com | |
| - www.euparliamentmonitor.com |
|
@copilot apply changes based on the comments in this thread |
…orkflows Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/8d6a2036-9dbf-4c48-8b3b-0e15162f5a8a Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
Applied the review feedback in commit Regarding the |
| - api.github.com | ||
| - data.europarl.europa.eu | ||
| - "*.europa.eu" | ||
| - hack23.com | ||
| - www.hack23.com | ||
| - riksdagsmonitor.com | ||
| - www.riksdagsmonitor.com | ||
| - euparliamentmonitor.com | ||
| - www.euparliamentmonitor.com |
There was a problem hiding this comment.
These updates are only in the gh-aw markdown source. The compiled workflow that actually runs in GitHub Actions is news-breaking.lock.yml, and it still has the old firewall allowlist (no hack23.com / euparliamentmonitor.com) and no repo-memory tool config. Please run gh aw compile for the updated .md workflows and commit the regenerated *.lock.yml files in this PR; otherwise the new allowlists + repo-memory won’t take effect (and links may still be blocked by AWF).
|
@copilot apply changes based on the comments in this thread |
Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/730e0e23-5283-4b0f-9f46-9c4681a49437 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
Ran |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
.github/workflows/news-week-ahead.md
Outdated
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
.github/workflows/news-breaking.md
Outdated
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
.github/workflows/news-motions.md
Outdated
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat commands will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to maintain editorial context across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json 2>/dev/null || echo '[]' | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' |
.github/workflows/news-translate.md
Outdated
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to track translation progress across runs. | ||
|
|
||
| **At workflow START** — read prior context: | ||
| ```bash | ||
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/translation-log.json 2>/dev/null || echo '[]' |
There was a problem hiding this comment.
Repo-memory path in the instructions doesn't match the actual directory used by the compiled workflow (lock file uses /tmp/gh-aw/repo-memory/default). With the current /tmp/gh-aw/repo-memory-default/ path, the START cat command will always miss prior context. Update the path references in this section to the correct repo-memory directory.
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory-default/`. Use it to track translation progress across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory-default/memory/news-generation/translation-log.json 2>/dev/null || echo '[]' | |
| This workflow has access to **persistent repo memory** at `/tmp/gh-aw/repo-memory/default`. Use it to track translation progress across runs. | |
| **At workflow START** — read prior context: | |
| ```bash | |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/translation-log.json 2>/dev/null || echo '[]' |
|
@copilot apply changes based on the comments in this thread |
… /tmp/gh-aw/repo-memory/default/ in all 10 workflows Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/521e808c-c96f-4ee9-92fd-c8d56875caa3 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
Fixed in |
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | |
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. | |
| **Scope clarification**: The **`news/`-only file creation rule applies only to the main checked-out repository workspace**. It does **not** prohibit updating the separate repo-memory workspace under `/tmp/gh-aw/repo-memory/default/memory/news-generation/`, which is explicitly allowed for the memory files below. | |
| 1. **`/tmp/gh-aw/repo-memory/default/memory/news-generation/article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | |
| 2. **`/tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. |
| cat /tmp/gh-aw/repo-memory/default/memory/news-generation/editorial-context.md 2>/dev/null || echo 'No prior context' | ||
| ``` | ||
|
|
||
| **At workflow END** — update memory (keep concise, max 50KB per file): |
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| **At workflow END** — update memory (keep concise, max 50KB per file): | |
| **At workflow END** — update memory (keep concise, max 50KB per file): | |
| > Clarification: the workflow rule to only create files under `news/` applies to the **main repository workspace**. Writing the repo-memory files under `/tmp/gh-aw/repo-memory/default/memory/news-generation/` is **explicitly allowed** for this workflow and does not count as creating files outside `news/` in the checked-out repository. |
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this month. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this month. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`article-log.json`** — Append today's generated article metadata (date, type, slug, headline, key topics). Keep last 30 entries. | ||
| 2. **`editorial-context.md`** — Brief summary of today's key findings, ongoing stories to track, and topics already covered this week. | ||
|
|
There was a problem hiding this comment.
The workflow scope restriction says the agent must only create files under news/, but this new section asks the agent to update repo-memory files at the end of the run. Please clarify that the news/-only rule applies to the main workspace, and that writing under /tmp/gh-aw/repo-memory/default/memory/news-generation/ (repo-memory branch) is explicitly allowed.
| **At workflow END** — update memory (keep concise, max 50KB per file): | ||
| 1. **`translation-log.json`** — Append today's translation metadata (date, source article, target languages, status). Keep last 30 entries. | ||
|
|
There was a problem hiding this comment.
This workflow states it must only create translated files under news/, but this new section also asks the agent to update repo-memory (/tmp/gh-aw/repo-memory/default/memory/news-generation/translation-log.json). Please clarify that writing to repo-memory is allowed even though the main workspace must remain news/-only.
|
@copilot apply changes based on the comments in this thread |
…mpile lock files Agent-Logs-Url: https://github.com/Hack23/euparliamentmonitor/sessions/d0386a11-2b83-480c-bf33-132a217407c3 Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>
Added scope clarification in |
euparliamentmonitor.com+www.euparliamentmonitor.comto network allowlists in all 10 news workflow .md fileseuparliamentmonitor.com+www.euparliamentmonitor.comto safe-outputs allowed-domains in all 10 news workflow .md files.github/skills/gh-aw-firewall.mdwith euparliamentmonitor.com documentation.github/skills/github-agentic-workflows.mdexamples with euparliamentmonitor.comanalysis/methodologies/ai-driven-analysis-guide.mdHack23 ecosystem cross-reference table.lock.ymlfiles viagh aw compilewith shared concurrency group patch/tmp/gh-aw/repo-memory-default/to/tmp/gh-aw/repo-memory/default/in all 10 workflowsnews/-only rule applies to main workspace, repo-memory writes are explicitly allowed