Skip to content
cccskills
FreePublished on GitHub

context-sync

Use this skill to check whether repository AGENTS.md context files have fallen behind code changes. It compares tracked file ownership with Git revisions, prepares focused update bundles for affected layers, and advances sync markers after the context is updated.

OriginalShow the original description

Sync AGENTS.md context files with the current codebase. Use whenever you want to check if context is stale, update context after code changes, or regenerate a specific layer's AGENTS.md. Triggers on: 'sync context', 'update context', 'is context stale', 'refresh AGENTS.md', 'context out of date', 'update the context files', or any time code has changed and you want the AGENTS.md files to reflect it.

flexprice/flexprice6,876 stars(GitHub 6,876 + cccskills 0)

Run score Not tested

See how to install

Good for

  • Checking whether AGENTS.md files are stale
  • Refreshing context after code changes
  • Updating one layer’s context file
  • Preparing a focused context update

Files included (3)

  • SKILL.md3.6 KB
  • ci-job.yml1.1 KB
  • sync.py8.0 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

Context Sync Skill

Keeps all Plane-2 AGENTS.md files in sync with the codebase using git-SHA watermarks. Run this any time after code changes to check for and fix stale context.

What this skill does

  1. Reads .context/manifest.yaml — the registry of all context nodes and their synced_sha.
  2. For each node: diffs synced_sha..HEAD and intersects with that node's owns globs.
  3. Nodes with no owned changes → SHA bumped automatically (no LLM work needed).
  4. Nodes with owned changes → prints the current AGENTS.md + scoped diff as a prompt bundle for the LLM to update minimally.
  5. After updates, advances synced_sha in the manifest.

How to run

Step 1 — Check what's stale

cd /path/to/flexprice  # repo root
pip install pyyaml --break-system-packages -q
python3 .context/sync/sync.py --check

Exits 0 if all current. Exits 1 and lists stale nodes if any are behind HEAD.

Step 2 — Sync (generate update bundles)

python3 .context/sync/sync.py --sync
  • Nodes with no owned changes: SHA bumped automatically, no output.
  • Nodes with changes: prints a prompt bundle (current AGENTS.md + scoped diff + instructions).

Step 3 — Update each stale AGENTS.md

For each printed bundle, the LLM (this session) reads the bundle and produces an updated AGENTS.md. Rules for the update:

  • Update "Key files" table if files were added, removed, or renamed.
  • Update "Patterns" or "Common pitfalls" if the diff reveals new patterns.
  • Do NOT rewrite sections that haven't changed.
  • Do NOT add improvement suggestions — those go in .context/findings/<layer>.md.
  • Bump synced_sha to current HEAD and synced_at to now in frontmatter.

Step 4 — Advance the manifest

After saving each updated file:

python3 .context/sync/sync.py --advance --file internal/service/AGENTS.md
# repeat for each updated file

Step 5 — Verify

python3 .context/sync/sync.py --check
# Should print: All context nodes are current.

Sync a single node only

python3 .context/sync/sync.py --sync --file internal/service/AGENTS.md
python3 .context/sync/sync.py --advance --file internal/service/AGENTS.md

Adding a new layer to the registry

When you create a new AGENTS.md for a directory, register it:

# .context/manifest.yaml — add a new node:
- file: internal/repository/AGENTS.md
  owns:
    - "internal/repository/**"
  synced_sha: <current HEAD SHA from: git rev-parse HEAD>
  synced_at: <ISO 8601 now>

What NOT to do

  • Do not rewrite the entire AGENTS.md on every sync — minimal diffs only.
  • Do not add critique, TODOs, or "areas to improve" inside AGENTS.md — put those in .context/findings/.
  • Do not run --sync and skip --advance — the manifest won't update and the node stays stale.

Full example session

You: sync context
Claude:
  → runs python3 .context/sync/sync.py --check
  → finds internal/service/AGENTS.md is stale (3 files changed)
  → runs --sync, reads the prompt bundle
  → produces updated internal/service/AGENTS.md (minimal changes only)
  → runs --advance --file internal/service/AGENTS.md
  → runs --check again → "All context nodes are current."

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

apitest

FreeSummary

Run local HTTP API checks with a staged workflow: confirm the environment, gather credentials interactively, probe endpoints with curl, and record status codes and response excerpts. For large suites, it can divide independent endpoint groups among workers and merge their results.

  • Smoke-checking a local API
  • Testing authenticated endpoints with curl
  • Organizing checks across many endpoints
flexprice/flexprice6.9KUpdated Oct 29, 2026

arch

FreeSummary

Navigate FlexPrice’s codebase using its architecture maps, dependency graphs, hotspot notes, and flow documents. Trace a feature across layers and async systems, assess likely change impact, and identify documentation that should be updated after structural edits.

  • Getting oriented in the FlexPrice codebase
  • Tracing a feature across code layers
  • Assessing impact before a refactor
flexprice/flexprice6.9KUpdated Oct 29, 2026

compose

FreeSummary

Use this skill to set up or manage FlexPrice’s local Docker environment. It provides commands to start and stop services, launch Kafka UI, and follow application logs while troubleshooting.

  • Setting up the local development environment
  • Starting or stopping FlexPrice services
  • Opening Kafka UI during development
flexprice/flexprice6.9KUpdated Oct 29, 2026

devenv

FreeSummary

Plan a local development setup by mapping which services run in Docker Compose and which use hosted endpoints such as RDS or managed Kafka. The skill guides environment configuration, service checks, migrations, and choosing an application mode.

  • Preparing a hybrid local and hosted development setup
  • Starting an API, consumer, or Temporal worker locally
  • Checking environment settings before migrations
flexprice/flexprice6.9KUpdated Oct 29, 2026

gh

FreeSummary

Use the GitHub CLI to review pull requests and issues, inspect CI check results and failed logs, and prepare a pull request with a summary and test plan. The skill also outlines the repository review steps before creation.

  • Reviewing a pull request or issue from the CLI
  • Finding the cause of a failed CI check
  • Preparing and creating a pull request
flexprice/flexprice6.9KUpdated Oct 29, 2026

godev

FreeSummary

Format Go code, check it with go vet, and run focused race tests or the full project test suite. The skill guides a repeatable verification flow after code changes, using Makefile targets when available.

  • Checking Go code after a change
  • Running race tests on edited packages
  • Running the full test suite
flexprice/flexprice6.9KUpdated Oct 29, 2026

See all skills by flexprice

Report a problem with this skill