update-context
/update-context
Use when the user says "remember this", "save that", "update the context" or "add this to the glossary", or a fact about the company, users, metrics, stakeholders or constraints should be written down now. Proposes small, specific edits to the context files and decision log and applies only approved ones. For an end-of-session retrospective, use done.
- Author
- Mosin112 ↗
- Repository
- pm-harness ↗
- Licence
- MIT
- Last repository change
- 12.11.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 12.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Use when the user says "remember this", "save that", "update the context" or "add this to the glossary", or a fact about the company, users, metrics, stakeholders or constraints should be written down now. Proposes small, specific edits to the context files and decision log and applies only approved ones. For an end-of-session retrospective, use done.
Triggers from the description
Original author description
Use when the user says "remember this", "save that", "update the context" or "add this to the glossary", or a fact about the company, users, metrics, stakeholders or constraints should be written down now. Proposes small, specific edits to the context files and decision log and applies only approved ones. For an end-of-session retrospective, use done.
How it works
- Needs
- Steps
- Decision log entry format
SKILL.md preview
--- name: update-context description: Use when the user says "remember this", "save that", "update the context" or "add this to the glossary", or a fact about the company, users, metrics, stakeholders or constraints should be written down now. Proposes small, specific edits to the context files and decision log and applies only approved ones. For an end-of-session retrospective, use done. --- # Update the context library The harness gets better only if what it learns is written down. This skill is the memory layer: it turns a session's lessons into durable, reviewable edits. ## Needs Before starting, run `python3 .pm-harness/pm_check.py needs update-context` and confirm each tool answers (preflight in `../../reference/connectors.md`). - `docs`, `team-chat` or `meeting-notes` · optional · checking a fact against its source before it is written down ## Steps 1. Review the session: decisions made, facts learned about users, metrics, stakeholders or constraints, corrections the PM made to your output, and repeated instructions the PM had to give you. 2. For each lesson, pick the one file it belongs in (see the routing index in `AGENTS.md`). Decisions go to `decisions/decision-log.md`; new terms to `context/glossary.md`; anything engineering asked for at handoff to `context/definition-of-done.md`. A correction to how a skill behaves goes to `context/writing-style.md` or is reported as a suggested skill change. 3. Present the proposals as a numbered list. For each: the file, the exact text to add or replace, one line on why, and where the fact came from (a doc, a thread, a meeting). When a fact can be checked against its source with the tools above, check it first. A new or changed tool goes to `context/tools.md` as a registry row (need, tool, role, source, trusted for). Keep each edit small. 4. **Wait for approval.** Apply only the items the PM approves, exactly as approved. Never rewrite a context file silently, and never delete content without explicit approval. 5. If a file now exceeds ~150 lines, propose a split and a routing-index update. To look for patterns across several sessions rather than this one, use the `done` skill. ## Decision log entry format ``` ## YYYY-MM-DD: <decision in one line> - **Context:** why this came up (link to item, meeting or signal) - **Decision:** what was decided, by whom - **Alternatives considered:** … - **Revisit if:** the condition that would reopen it ```
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · Mosin112/pm-harness | 100 | MIT |
| ducky8x/session-kit | 86 | MIT |
Database snapshot: 13.11.2026 · package version 5b1d9083a100 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100done
Use when the PM says "done", "wrap up", "close the session", "what did we learn" or "what should we learn from this", or asks for a session retrospective. Reviews this session and recent ones for repeated corrections, re-explained facts and friction, then proposes context… - 100stakeholder-update
Use when the user wants a status update, weekly update, roadmap update or progress summary for stakeholders or leadership. Drafts a short update from the tracker, work items and decision log, leading with what the reader needs to do. - 100startup-context
When the user wants to create or update their startup context document. Also use when the user mentions "set up context", "tell you about my startup", or any other skill needs context that doesn't exist yet. - 100context-freshness
Monthly judgment pass on context: files past 90 days, contradictions between strategy, knowledge and the decision log, which refresh to run. Use when "is our context current", or monthly. - 90session-reconcile
Check what a session changed against docs, agent instructions and memory; list each mismatch, fix only approved ones. USE WHEN records may lag reality. NOT FOR handoff notes or progress files.