docs: manual test plan + full test-run results (10 findings)#78
Merged
Conversation
Full execution of docs/manual-test.md against agent.zeroshot.lol. 10 findings recorded with severity and fix direction, notably: - P0 DKIM key on disk does not match DNS TXT (root cause of outbound dkim=fail at Gmail) - P0 Shell injection in on_receive cmd template expansion - P1 MCP write ops (email_mark_read, etc.) fail when MCP runs as non-root due to root:root 0644 mailbox files
…lans Code-review-backed fix plans for each of the 10 findings, with file:line refs, effort estimates, and a priority order. No code changes yet — this consolidates the investigation so fixes can be sequenced. - #10 DKIM mismatch: not a code bug; DNS republish + optional startup check - #9 shell injection: pass trigger vars via env, not string substitution - #8 MCP writes: route state mutations through daemon UDS - #7 claude-code hint: print `claude mcp add` command post-install - #4 send config read: move mailbox resolution to daemon side - #2 SPF: plumb envelope MAIL FROM from smtp session through ingest - #5 wildcard send: remove wildcard branch from resolve_from_mailbox - #1 mailbox create: add restart hint (or route via daemon) - #3 plan wording: clarify "compose new" in docs/manual-test.md
uzyn
added a commit
that referenced
this pull request
Apr 21, 2026
* docs: add manual test results with findings Full execution of docs/manual-test.md against agent.zeroshot.lol. 10 findings recorded with severity and fix direction, notably: - P0 DKIM key on disk does not match DNS TXT (root cause of outbound dkim=fail at Gmail) - P0 Shell injection in on_receive cmd template expansion - P1 MCP write ops (email_mark_read, etc.) fail when MCP runs as non-root due to root:root 0644 mailbox files * docs: add Recommended fixes section with per-finding implementation plans Code-review-backed fix plans for each of the 10 findings, with file:line refs, effort estimates, and a priority order. No code changes yet — this consolidates the investigation so fixes can be sequenced. - #10 DKIM mismatch: not a code bug; DNS republish + optional startup check - #9 shell injection: pass trigger vars via env, not string substitution - #8 MCP writes: route state mutations through daemon UDS - #7 claude-code hint: print `claude mcp add` command post-install - #4 send config read: move mailbox resolution to daemon side - #2 SPF: plumb envelope MAIL FROM from smtp session through ingest - #5 wildcard send: remove wildcard branch from resolve_from_mailbox - #1 mailbox create: add restart hint (or route via daemon) - #3 plan wording: clarify "compose new" in docs/manual-test.md
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
docs/manual-test.md(T0–T13) end-to-end againstagent.zeroshot.loland recorded each acceptance bullet's PASS / FAIL / DEVIATION with evidence indocs/manual-test-results.md.Top findings
dkim._domainkey.agent.zeroshot.lol. Every outbound signature fails at receivers (root cause of the Gmail-sidedkim=failseen in T4/T6/T7).on_receive cmdtemplate expansion is a shell-injection vector; the plan's own example breaks on realisticName <addr>senders.email_mark_read, …) fail as EACCES when MCP runs as a non-root user because mailbox files areroot:root 0644.aimx agent-setup claude-codedoesn't register the MCP server;claude -pcan't see the tool until manualclaude mcp add.aimx sendas a non-root user fails to open/etc/aimx/config.toml(0640 root:root)."none"for mail from a domain that clearly publishes SPF covering the Received-IP.aimx sendaccepts anyFrom:under the operator's domain when a wildcard catchall exists.aimx mailbox createsilently misroutes mail until the daemon is restarted; no restart hint printed.Findings #3 and #6 are cross-referenced in the doc (#6 is the externally-observed symptom of #10; #3 is a plan-wording nit about forwarded messages).
Deviations (non-defects)
[LOCAL]tests (T6, T7) ran on the VPS becauseclaude+codexare installed there..claude-plugin/plugin.json;~/.codex/skills/aimx/vs~/.codex/plugins/aimx/).Side-effects left on the VPS
/etc/aimx/config.tomlmode0640→0644(workaround for finding [Sprint 3] MCP Server #4).[mailboxes.test]has a trust + on_receive trigger config from T8./tmp/aimx-trigger-*.flagartifacts from T8/T9 still present; clean withsudo rm /tmp/aimx-trigger*.flag /tmp/aimx-trigger.log.Test plan
docs/manual-test-results.mdend-to-end — each test section maps 1:1 todocs/manual-test.mdacceptance bullets.dig +short TXT dkim._domainkey.agent.zeroshot.loland comparing to/etc/aimx/dkim/public.key.