memory-export-summary
Skill1 starsChanged 4 months agoMITView on GitHub
Compress a PM workflow project's current state into a concise memory-safe summary with continuation pointers instead of copying full project assets into memory. Use when Codex needs to write durable memory notes, export project status for later continuation, or summarize the current baseline/version/stage without duplicating the full requirement/demo/PRD asset set.
In this repository17 other files it gives its agents.
- audience-judgment-framingSkill
- embedded-prd-auditSkill
- freeze-readiness-checkSkill
- pm-demo-designSkill
- pm-design-absorptionSkill
- pm-devil-advocateSkill
- pm-embedded-prdSkill
- pm-prd-writerSkill
- pm-project-opsSkill
- pm-react-prototype-executionSkill
- pm-requirement-intakeSkill
- pm-requirement-reverse-auditSkill
- pm-value-anchorSkill
- prototype-export-intakeSkill
- request-triageSkill
- rule-change-signalSkill
- state-auditSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
24 rules. About 0 are common, 23 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈23
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
Use this skill to make memory useful for navigation and continuity instead of turning it into a duplicate project archive.
- Distinctive: line 34
If the current project state is too unstable to summarize as durable memory, say so and export only the minimal continuation pointer.
- Distinctive: line 48
Prefer project-ops outputs and current baseline files over long chat history.
- Distinctive: line 80
Do not overstate trust. If the project is still shifting, say so.
- Distinctive: line 84
Prefer facts such as:
- Distinctive: line 93
Avoid pulling in long explanations unless they are truly needed for later continuity.
- Distinctive: line 97
When necessary, split the export into:
- Distinctive: line 102
If the unstable part is large, keep it very short.
- Distinctive: line 106
Always include the best continuation references available, such as:
- Distinctive: line 121
If a detail already exists in project assets, prefer a pointer over duplication.
- Distinctive: line 125
Prefer this compact structure:
- Distinctive: line 138
Keep it concise. The most important fields are:
- Distinctive: line 185
Use these tests when unsure.
- Distinctive: line 191
If yes, prefer a pointer.
- Distinctive: line 197
If it is draft noise, exclude or compress it heavily.
- Distinctive: line 201
If future-you read only this summary, would they know what is current and where to resume?
- Distinctive: line 203
If no, improve the baseline and continuation lines rather than adding raw detail.
- Distinctive: line 209
If no, downgrade the confidence and export level.
- Very specific: line 223
every page/module/state description
- Distinctive: line 239
Read these only when useful:
- Distinctive: line 246
When the project is still unstable:
- Distinctive: line 247
prefer a minimal or working summary export
- Distinctive: line 250
When a baseline is trustworthy:
- Distinctive: line 252
keep the rest short and pointer-based
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Memory Export Summary
Act as a command-style skill for turning project assets into concise memory-safe continuation notes.
Core job
Take the current project state and produce a compact summary that:
- captures the durable facts worth remembering
- points to the current baseline and continuation entry
- avoids copying large project assets into memory
- distinguishes stable truth from temporary working detail
Use this skill to make memory useful for navigation and continuity instead of turning it into a duplicate project archive.
Boundaries
This skill is responsible for:
- summarizing durable project state
- naming the current version/stage/baseline clearly
- pointing to the best continuation paths
- separating stable conclusions from working detail
This skill is not responsible for:
- storing full rules, full demos, or full PRDs in memory
- replacing project-ops asset management
- acting as the primary source of truth for the project
- preserving every draft detail or chat fragment
If the current project state is too unstable to summarize as durable memory, say so and export only the minimal continuation pointer.
Inputs
Work from whatever current project evidence exists, such as:
- project slug or name
- current version/stage
- current baseline set
- freeze status
- core rule/model conclusion
- current demo/PRD status
- key unresolved items
- continuation entry files or project paths
Prefer project-ops outputs and current baseline files over long chat history.
Core memory principle
Memory should store:
- what this project currently is
- what round it is in
- what the current trusted baseline is
- what the main durable conclusion is
- where to continue next
Memory should not store:
- the full rule document
- the full demo description
- the full embedded PRD
- every version diff
- raw case piles that already live in project assets
In short:
- memory = summary + pointer
- project assets = full detail
Working flow
1) Identify the current trust level
First judge what kind of state you are exporting:
- working draft only
- review candidate baseline
- frozen delivery baseline
- historical summary
Do not overstate trust. If the project is still shifting, say so.
2) Extract only durable facts
Prefer facts such as:
- project identity
- current effective version/stage
- current baseline status
- durable rule/model conclusion
- major current scope
- major open issue if it materially affects continuation
- best continuation entry
Avoid pulling in long explanations unless they are truly needed for later continuity.
3) Separate stable from unstable
When necessary, split the export into:
- stable summary
- unresolved / still-moving items
- continuation pointers
If the unstable part is large, keep it very short.
4) Point to the real assets
Always include the best continuation references available, such as:
- project root
- current baseline manifest
- current version directory
- continuation-entry file
- embedded PRD path if relevant
The point of memory is to help future work re-enter the project quickly.
5) Keep the summary compact
Default target:
- short enough to drop into memory cleanly
- rich enough that future-you knows what is current and where to resume
If a detail already exists in project assets, prefer a pointer over duplication.
Recommended output format
Prefer this compact structure:
## Project memory summary
- Project:
- Current stage:
- Current version:
- Current baseline:
- Durable conclusion:
- Open issue / caution:
- Continue from:Keep it concise. The most important fields are:
- current stage
- current version
- durable conclusion
- continue from
Export levels
Level 1: Minimal pointer export
Use when:
- the project is still unstable
- there is not yet a trustworthy durable conclusion
Output only:
- project identity
- current stage/version if known
- main continuation pointer
- short caution
Level 2: Working summary export
Use when:
- the round is coherent enough to summarize
- but still not a strong frozen baseline
Output:
- current stage/version
- current main direction
- key unresolved item(s)
- continuation pointer
Level 3: Baseline summary export
Use when:
- there is a review-grade or frozen baseline worth remembering durably
Output:
- stage/version
- current baseline set
- stable rule/model conclusion
- main scope
- continuation pointer
- one short caution if needed
Decision heuristics
Use these tests when unsure.
Test 1: Already-lives-in-assets test
Ask:
- Does this detail already live safely in project files and not need duplication?
If yes, prefer a pointer.
Test 2: Durable-next-week test
Ask:
- Will this still help next week, or is it just today's draft noise?
If it is draft noise, exclude or compress it heavily.
Test 3: Re-entry test
Ask:
- If future-you read only this summary, would they know what is current and where to resume?
If no, improve the baseline and continuation lines rather than adding raw detail.
Test 4: Trustfulness test
Ask:
- Does this summary honestly reflect how stable the current state is?
If no, downgrade the confidence and export level.
Typical export contents worth keeping
Usually worth keeping:
- current main version/stage
- current trusted baseline identity
- main rule or model conclusion if it has stabilized
- whether the project is at draft/review/frozen stage
- major unresolved blocker if it affects continuation
- best file/path to resume from
Usually not worth copying in full:
- complete rule list
- every page/module/state description
- full linkage map
- large diff lists
- raw review chatter
Anti-patterns
Avoid these mistakes:
- copying the whole PRD into memory
- copying the whole demo description into memory
- turning memory into a duplicate project filesystem
- exporting unstable exploratory conclusions as durable truth
- forgetting to include the best continuation path
References to load when needed
Read these only when useful:
../pm-project-ops/references/project-structure-spec.mdfor likely project entry and asset locations../pm-project-ops/references/version-and-snapshot-rules.mdfor stage/version/baseline discipline../freeze-readiness-check/SKILL.mdwhen current stability is unclear and trust level must be judged before export
Handoff expectations
When the project is still unstable:
- prefer a minimal or working summary export
- point strongly to the continuation entry
When a baseline is trustworthy:
- export the durable conclusion and baseline identity clearly
- keep the rest short and pointer-based
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

