framework-resume
Skill1 starsChanged 53 days agoMITView on GitHub
Safely resume framework work by reading the latest Checkpoint, validating repository and Artifact state, detecting drift, revalidating affected work, and recomputing the Minimum Necessary Path. Use after session restart, context loss, pause, or handoff.
In this repository21 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- adaptive-testingSkill
- architecture-decisionSkill
- artifact-validationSkill
- change-impactSkill
- clarification-gateSkill
- code-reviewSkill
- component-planningSkill
- failure-diagnosisSkill
- failure-escalationSkill
- framework-checkpointSkill
- gap-analysisSkill
- implement-componentSkill
- integration-gateSkill
- minimum-pathSkill
- project-intakeSkill
- quality-gateSkill
- reverse-engineeringSkill
- stage-gateSkill
- traceabilitySkill
- workflow-assessmentSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 7 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Run at session recovery, after PAUSED_AT_STAGE, after a handoff/context reset, or whenever repository activity may have occurred since the …
- Distinctive: line 25
Do not mutate project work until reconciliation identifies a safe next action.
- Distinctive: line 29
Read Project Configuration, latest Checkpoint, linked Artifacts, State, and target.
- Very specific: line 34
Run Change Impact for every material drift and mark affected Artifacts STALE/UNVERIFIED and Components UNVERIFIED as appropriate.
- Distinctive: line 39
Write a reconciled Checkpoint before continuing substantial work.
- Distinctive: line 44
Do not rerun completed work whose evidence remains current and sufficient.
- Distinctive: line 45
Do not continue a previously ready node if its dependencies changed.
- Distinctive: line 58
No trustworthy Checkpoint exists.
- Distinctive: line 65
Run Project Intake when no usable snapshot exists.
- Very specific: line 69
Stop when current state is reconciled and a safe next action is identified, when the target remains intentionally paused, or when …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Framework Resume
Purpose
Continue from persisted evidence without blindly trusting an old Checkpoint or hidden conversation history.
Trigger
Run at session recovery, after PAUSED_AT_STAGE, after a handoff/context reset, or whenever repository activity may have occurred since the last Checkpoint.
Inputs
config/project.md, user request, latest Checkpoint, State, Artifact Registry, and dependency/traceability records.- Current repository revision, branch/worktree, working-tree status, files, and available validation evidence.
- Open clarifications, decisions, assumptions, risks, and prior Minimum Necessary Path.
Prerequisites
- Locate the latest relevant Checkpoint by ID/time/repository lineage, not filename guess alone.
- Do not mutate project work until reconciliation identifies a safe next action.
Procedure
- Read Project Configuration, latest Checkpoint, linked Artifacts, State, and target.
- Inspect current repository revision, branch/worktree, dirty state, and relevant file changes.
- Compare repository/Artifact/Component revisions with the Checkpoint.
- Detect added, removed, modified, or externally changed assets since the snapshot.
- Reopen unresolved Clarifications, assumptions, decisions, risks, failures, and blockers.
- Run Change Impact for every material drift and mark affected Artifacts
STALE/UNVERIFIEDand ComponentsUNVERIFIEDas appropriate. - Revalidate affected Artifacts and current Gate evidence; preserve unaffected evidence with rationale.
- Reassess Workflow Level if scope/risk changed.
- Rerun Gap Analysis and recompute the Minimum Necessary Path from current evidence.
- Select the next dependency-ready action or report the exact blocker.
- Write a reconciled Checkpoint before continuing substantial work.
Decision Rules
- Repository evidence and exact revisions outrank stale Checkpoint claims, but conflicts must be investigated rather than silently overwritten.
- Do not rerun completed work whose evidence remains current and sufficient.
- Do not continue a previously ready node if its dependencies changed.
- A prior PASS is invalid when its evaluated revision or basis changed.
- Reaching a previous Target Stage remains paused unless the user requests a new target.
Outputs
- Resume/Reconciliation Report with Checkpoint and current repository comparison.
- Changed-since-checkpoint set and resulting state invalidations.
- Revalidation results, preserved evidence rationale, updated risks, and reconciled Checkpoint.
- Recomputed Minimum Necessary Path and next ready action/blocker.
Failure Cases
- No trustworthy Checkpoint exists.
- Multiple divergent worktrees/revisions claim continuity.
- Referenced Artifacts/evidence are missing.
- Repository contains unattributed or conflicting changes.
Escalation
Run Project Intake when no usable snapshot exists. Use Clarification Gate for ownership/intent conflicts, Failure Diagnosis for unexplained drift, and request explicit merge/rollback authority for divergent work.
Stop Conditions
Stop when current state is reconciled and a safe next action is identified, when the target remains intentionally paused, or when conflict/access/authority blocks recovery. Never continue solely because the old Checkpoint said a node was ready.
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.

