feat: Codex-style handoff prefix for compressed context summaries by teknium1 · Pull Request #916 · NousResearch/hermes-agent

teknium1 · 2026-03-11T12:41:26Z

Summary

Replaces the old [CONTEXT SUMMARY]: prefix on compressed summaries with a Codex-inspired handoff framing that tells the model what happened and how to use the summary.

What exactly changes

Before — summaries were inserted as:

[CONTEXT SUMMARY]: The user asked about X. The assistant searched for Y...

After — summaries are inserted as:

[CONTEXT COMPACTION] An earlier part of this conversation was summarized
to preserve context space. Below is the summary — use it to build on
the work already done and avoid duplicating effort:

The user asked about X. The assistant searched for Y...

Detailed changes

`agent/context_compressor.py`

New SUMMARY_PREFIX constant (module-level) — the handoff text prepended to every summary
New LEGACY_SUMMARY_PREFIX constant — keeps "[CONTEXT SUMMARY]:" for backward compatibility detection
New _with_summary_prefix() static method — normalizes model output:
- Strips any legacy [CONTEXT SUMMARY]: prefix the summarization model may have produced
- Prepends the new SUMMARY_PREFIX
_call_summary_model() — now calls _with_summary_prefix() instead of manually checking/prepending [CONTEXT SUMMARY]:
System message annotation — the note appended to the system prompt on first compression updated from the generic "[Note: Some earlier conversation turns may be summarized...]" to "[Note: Some earlier conversation turns have been compacted into a handoff summary... Build on that summary rather than re-doing work.]"

Test updates

Tests now import and check against SUMMARY_PREFIX instead of hardcoded "CONTEXT SUMMARY" strings
tests/test_413_compression.py mock return values updated to use the new prefix format
All 31 existing tests pass unmodified in logic — only string matching updated

What does NOT change

The compression algorithm (positional protection, boundary alignment, middle-turn summarization)
The role alternation logic (summary role still adapts to avoid consecutive same-role messages)
The summarization model used or trigger thresholds
The compression prompt sent to the summarization model (that's a separate PR)

Why this is better

The old prefix gave the model zero context about what the summary block is. The new prefix:

Explains what happened — context was compacted
Instructs the model — build on prior work, don't duplicate effort
Has a detectable marker — [CONTEXT COMPACTION] for programmatic detection

Inspired by PR #776 by @kshitijk4poor and the research in #499.

@kshitijk4poor

Replace the old '[CONTEXT SUMMARY]:' prefix on compressed summaries with a Codex-inspired handoff framing that tells the model what happened and how to use the summary. What changes: 1. New SUMMARY_PREFIX constant — the text prepended to every compressed summary: [CONTEXT COMPACTION] An earlier part of this conversation was summarized to preserve context space. Below is the summary — use it to build on the work already done and avoid duplicating effort: 2. _with_summary_prefix() helper — normalizes model output by stripping any legacy '[CONTEXT SUMMARY]:' prefix the summarization model may have produced, then prepends the new SUMMARY_PREFIX. 3. System message annotation updated — the note appended to the system prompt on first compression now says 'compacted into a handoff summary' and instructs 'build on that summary rather than re-doing work' instead of the old generic note. Why this is better: The old prefix ('[CONTEXT SUMMARY]: <raw text>') gave the model no context about what the summary is or how to use it. The new prefix explicitly frames it as a context compaction event and instructs the model to build on prior work rather than re-doing it. This reduces redundant tool calls and file re-reads after compression. What does NOT change: - The compression algorithm (positional protection, boundary alignment) - The role alternation logic (summary role adapts to avoid consecutive same-role messages) - The summarization model or trigger thresholds - LEGACY_SUMMARY_PREFIX is exported for backward compatibility Inspired by PR #776 by @kshitijk4poor and the research in #499.

Adapt PR #916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path.

teknium1 · 2026-03-14T09:33:39Z

Merged via PR #1273. I adapted the handoff-prefix idea onto current main's call_llm-based context compressor and tightened the wording to be more explicit about earlier work and session state.

* feat: improve context compaction handoff summaries Adapt PR #916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path. * fix: clearer error when docker backend is unavailable * fix: preserve docker discovery in backend preflight Follow up on salvaged PR #940 by reusing find_docker() during the new availability check so non-PATH Docker Desktop installs still work. Add a regression test covering the resolved executable path. --------- Co-authored-by: aydnOktay <xaydinoktay@gmail.com>

* feat: improve context compaction handoff summaries Adapt PR #916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path. * fix: clearer error when docker backend is unavailable * fix: preserve docker discovery in backend preflight Follow up on salvaged PR #940 by reusing find_docker() during the new availability check so non-PATH Docker Desktop installs still work. Add a regression test covering the resolved executable path. * test: make gateway async tests xdist-safe Replace sync test usage of asyncio.get_event_loop().run_until_complete() with asyncio.run() so tests do not depend on an ambient current event loop. Also create the email disconnect poll task inside a running loop. This fixes xdist/CI failures where workers have no current loop in MainThread. --------- Co-authored-by: aydnOktay <xaydinoktay@gmail.com>

…alls (#1289) * feat: improve context compaction handoff summaries Adapt PR #916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path. * fix: clearer error when docker backend is unavailable * fix: preserve docker discovery in backend preflight Follow up on salvaged PR #940 by reusing find_docker() during the new availability check so non-PATH Docker Desktop installs still work. Add a regression test covering the resolved executable path. * test: make gateway async tests xdist-safe Replace sync test usage of asyncio.get_event_loop().run_until_complete() with asyncio.run() so tests do not depend on an ambient current event loop. Also create the email disconnect poll task inside a running loop. This fixes xdist/CI failures where workers have no current loop in MainThread. * feat(skills): add phone-calls skill for outbound AI voice calls Reformulated from core tool (PR #847 feedback) into a skill with a standalone helper script. No new dependencies — uses only Python stdlib. Two providers supported: - Bland.ai (default): simple setup, one API key - Vapi: flexible, better voice quality via ElevenLabs/Deepgram + Twilio Includes: - SKILL.md with full procedure, safety rules, provider docs, pitfalls - scripts/phone_call.py CLI helper (call, status, diagnose commands) * feat(skills): expand phone-calls into optional telephony skill Follow up on salvaged PR #965 by moving the capability into optional-skills and broadening it from outbound AI calling to a full telephony skill. Add Twilio number provisioning, env/state persistence, SMS/MMS, inbound SMS polling, Vapi import helpers, and a provider decision tree while keeping telephony out of core runtime code. * docs(skills): clarify Hermes TTS telephony workflow --------- Co-authored-by: aydnOktay <xaydinoktay@gmail.com> Co-authored-by: mormio <morganemoss@gmai.com>

Adapt PR NousResearch#916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path.

* feat: improve context compaction handoff summaries Adapt PR NousResearch#916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path. * fix: clearer error when docker backend is unavailable * fix: preserve docker discovery in backend preflight Follow up on salvaged PR NousResearch#940 by reusing find_docker() during the new availability check so non-PATH Docker Desktop installs still work. Add a regression test covering the resolved executable path. --------- Co-authored-by: aydnOktay <xaydinoktay@gmail.com>

* feat: improve context compaction handoff summaries Adapt PR NousResearch#916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path. * fix: clearer error when docker backend is unavailable * fix: preserve docker discovery in backend preflight Follow up on salvaged PR NousResearch#940 by reusing find_docker() during the new availability check so non-PATH Docker Desktop installs still work. Add a regression test covering the resolved executable path. * test: make gateway async tests xdist-safe Replace sync test usage of asyncio.get_event_loop().run_until_complete() with asyncio.run() so tests do not depend on an ambient current event loop. Also create the email disconnect poll task inside a running loop. This fixes xdist/CI failures where workers have no current loop in MainThread. --------- Co-authored-by: aydnOktay <xaydinoktay@gmail.com>

…alls (NousResearch#1289) * feat: improve context compaction handoff summaries Adapt PR NousResearch#916 onto current main by replacing the old context summary marker with a clearer handoff wrapper, updating the summarization prompt for resume-oriented summaries, and preserving the current call_llm-based compression path. * fix: clearer error when docker backend is unavailable * fix: preserve docker discovery in backend preflight Follow up on salvaged PR NousResearch#940 by reusing find_docker() during the new availability check so non-PATH Docker Desktop installs still work. Add a regression test covering the resolved executable path. * test: make gateway async tests xdist-safe Replace sync test usage of asyncio.get_event_loop().run_until_complete() with asyncio.run() so tests do not depend on an ambient current event loop. Also create the email disconnect poll task inside a running loop. This fixes xdist/CI failures where workers have no current loop in MainThread. * feat(skills): add phone-calls skill for outbound AI voice calls Reformulated from core tool (PR NousResearch#847 feedback) into a skill with a standalone helper script. No new dependencies — uses only Python stdlib. Two providers supported: - Bland.ai (default): simple setup, one API key - Vapi: flexible, better voice quality via ElevenLabs/Deepgram + Twilio Includes: - SKILL.md with full procedure, safety rules, provider docs, pitfalls - scripts/phone_call.py CLI helper (call, status, diagnose commands) * feat(skills): expand phone-calls into optional telephony skill Follow up on salvaged PR NousResearch#965 by moving the capability into optional-skills and broadening it from outbound AI calling to a full telephony skill. Add Twilio number provisioning, env/state persistence, SMS/MMS, inbound SMS polling, Vapi import helpers, and a provider decision tree while keeping telephony out of core runtime code. * docs(skills): clarify Hermes TTS telephony workflow --------- Co-authored-by: aydnOktay <xaydinoktay@gmail.com> Co-authored-by: mormio <morganemoss@gmai.com>