Context Catch-Up
Use this only at a fresh-session or true-resume boundary in a repository that
already has context-harness files.
Repeat catch-up only after a real resume/compaction boundary, a materially
changed user objective, or an external context change. Otherwise use loaded
context, and use context-maintain only for updates, closeout, lesson capture,
or plan stress-testing.
Compatibility Check
Before ordinary catch-up, quickly inspect AGENTS.md, CONTEXT.md, and
NOW.md for:
<!-- context-harness:schema v3 --> in generated context files.
- A small
AGENTS.md Context Contract plus generated Context Index.
- Required
CONTEXT.md sections: Project, Structure, Operating Constraints,
Workflow, Language, Relationships, Flagged Ambiguities, Learned Patterns.
- A non-empty
NOW.md with current focus, blockers, immediate next step, and
touched files.
If files are empty, partial, stale, or missing the v3 schema marker, report the
drift in the catch-up summary and route inspection or repair work to explicit
context-upgrade. Do not rewrite partial or schema-drifted layouts from
catch-up, and do not let non-blocking harness drift replace the user's requested
project task.
Read Order
- Read the small
NOW.md recovery packet first.
- If the user supplied an objective, task, question, or relevant context, use
that as the hydrate query. If not, derive the query from
NOW.md focus and
immediate next step.
- Reconcile the user's context with
NOW.md: prefer the current user objective
when scope changed, but surface material conflicts with repository state.
- Run
node scripts/context-index.js hydrate "<query>" before opening
PLAN.md, chunks, or bulky/task-specific context.
- Read concise
CONTEXT.md directly only while it fits the startup budget;
otherwise read hydrate-selected cards/sections. Even when it fits, keep the
catch-up selective and do not load unrelated project history.
- Use selected cards before large
PLAN.md sections, chunks, or raw sources.
- If there is no index, use only enough
CONTEXT.md and AGENTS.md fallback
context for the current objective and route repair to context-upgrade.
- If the project uses ADRs, skim only titles/dates relevant to the current task.
Output
Give the user a short catch-up summary:
- Current focus
- Active blockers
- Immediate next step
- Operating constraints or terms that matter for the requested task
- Any uncertainty that should be resolved before editing
Guardrails
- Do not rewrite context files during catch-up except to capture a durable
correction from the user; route layout repair to explicit
context-upgrade.
- If
hydrate, check, or generated indexes fail or warn about stale generated
files, mention the drift and use current markdown fallback context. Keep index
repair as a follow-up unless the drift blocks correctness or safety.
- Do not summarize raw external content into
CONTEXT.md; route task-local
discoveries to PLAN.md when work begins.
- Keep the summary short enough that it helps the next action, not replaces it.
1---2name: context-catch-up3description: Catch up a fresh agent session or true resumed session from context-harness files before planning or editing; do not use for ordinary mid-session follow-up turns after context is already loaded.4---56# Context Catch-Up78Use this only at a fresh-session or true-resume boundary in a repository that9already has context-harness files.1011Repeat catch-up only after a real resume/compaction boundary, a materially12changed user objective, or an external context change. Otherwise use loaded13context, and use `context-maintain` only for updates, closeout, lesson capture,14or plan stress-testing.1516## Compatibility Check1718Before ordinary catch-up, quickly inspect `AGENTS.md`, `CONTEXT.md`, and19`NOW.md` for:2021- `<!-- context-harness:schema v3 -->` in generated context files.22- A small `AGENTS.md` Context Contract plus generated Context Index.23- Required `CONTEXT.md` sections: Project, Structure, Operating Constraints,24 Workflow, Language, Relationships, Flagged Ambiguities, Learned Patterns.25- A non-empty `NOW.md` with current focus, blockers, immediate next step, and26 touched files.2728If files are empty, partial, stale, or missing the v3 schema marker, report the29drift in the catch-up summary and route inspection or repair work to explicit30`context-upgrade`. Do not rewrite partial or schema-drifted layouts from31catch-up, and do not let non-blocking harness drift replace the user's requested32project task.3334## Read Order35361. Read the small `NOW.md` recovery packet first.372. If the user supplied an objective, task, question, or relevant context, use38 that as the hydrate query. If not, derive the query from `NOW.md` focus and39 immediate next step.403. Reconcile the user's context with `NOW.md`: prefer the current user objective41 when scope changed, but surface material conflicts with repository state.424. Run `node scripts/context-index.js hydrate "<query>"` before opening43 `PLAN.md`, chunks, or bulky/task-specific context.445. Read concise `CONTEXT.md` directly only while it fits the startup budget;45 otherwise read hydrate-selected cards/sections. Even when it fits, keep the46 catch-up selective and do not load unrelated project history.476. Use selected cards before large `PLAN.md` sections, chunks, or raw sources.487. If there is no index, use only enough `CONTEXT.md` and `AGENTS.md` fallback49 context for the current objective and route repair to `context-upgrade`.508. If the project uses ADRs, skim only titles/dates relevant to the current task.5152## Output5354Give the user a short catch-up summary:5556- Current focus57- Active blockers58- Immediate next step59- Operating constraints or terms that matter for the requested task60- Any uncertainty that should be resolved before editing6162## Guardrails6364- Do not rewrite context files during catch-up except to capture a durable65 correction from the user; route layout repair to explicit `context-upgrade`.66- If `hydrate`, `check`, or generated indexes fail or warn about stale generated67 files, mention the drift and use current markdown fallback context. Keep index68 repair as a follow-up unless the drift blocks correctness or safety.69- Do not summarize raw external content into `CONTEXT.md`; route task-local70 discoveries to `PLAN.md` when work begins.71- Keep the summary short enough that it helps the next action, not replaces it.