agentleFS
loganware05loganware05/captains-compass-cursor/.cursor/skills/technology-intelligence-live/SKILL.md

technology-intelligence-live

Skill2 starsChanged 32 days agoMITView on GitHub

Opt-in live GitHub Stars Technology Intelligence for capability planning (Captain-gated)

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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
  1. Very specific:

    Verify gh auth status succeeds locally.

    line 51
  2. Distinctive:

    Use Technology Intelligence Candidates in capability plans as informational input only.

    line 61

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Technology Intelligence (Live GitHub Stars)

Use this Skill when

The Captain wants live discovery signals from starred GitHub repositories during capability planning — not offline fixtures and not CI defaults.

Prerequisites

  • gh CLI installed and authenticated (gh auth login)
  • Explicit opt-in: COMPASS_TI_PROVIDER=github-stars
  • Captain understands candidates are NOT APPROVED FOR EXECUTION

Inputs

  • Natural-language objective (same as capability planning)
  • Optional --top N for result limit

Topology (required)

TI refresh / learning CLIs live only in the control repository. Product / sandbox checkouts do not contain scripts/refresh-ti-cache.sh, scripts/run-skill-learning-loop.sh, or scripts/northstar.

CONTROL=/path/to/captains-compass-cursor
SANDBOX=/path/to/captain-compass-sandbox

Prefer:

"$CONTROL/scripts/northstar" skills refresh --repo "$SANDBOX"
"$CONTROL/scripts/northstar" skills learn --repo "$SANDBOX" --objective "…"

Equivalent:

"$CONTROL/scripts/refresh-ti-cache.sh" --repo-root "$SANDBOX"
"$CONTROL/scripts/run-skill-learning-loop.sh" --repo-root "$SANDBOX" --objective "…"

Procedure

  1. Verify gh auth status succeeds locally.
  2. Query live starred repos (read-only):

    COMPASS_TI_PROVIDER=github-stars \
      "$CONTROL/scripts/query-technology-intelligence.sh" --query "accessible react forms"
  3. Review JSON candidates — each has approved_for_execution: false and lifecycle DISCOVERED.
  4. Use Technology Intelligence Candidates in capability plans as informational input only; never clone or execute starred repos without Captain approval.
  5. To promote a candidate, switch to Skill candidate-promotion (staging ceiling SANDBOX_TESTED; Skill install requires Captain PR).
  6. Offline cache (optional — separate provider):

    "$CONTROL/scripts/northstar" skills refresh --repo "$SANDBOX"
    "$CONTROL/scripts/northstar" skills refresh --repo "$SANDBOX" --if-stale 24
    # equivalent:
    "$CONTROL/scripts/refresh-ti-cache.sh" --repo-root "$SANDBOX"
    "$CONTROL/scripts/refresh-ti-cache.sh" --repo-root "$SANDBOX" --if-stale 24
    COMPASS_TI_PROVIDER=github-stars-cached \
      "$CONTROL/scripts/query-technology-intelligence.sh" --query "accessible react forms"

    Cache envelope includes fetched_at (and legacy refreshed_at).

  7. Hugging Face file TI (offline fixtures / Captain export dir — no Hub network in CI):

    COMPASS_TI_PROVIDER=huggingface-file \
      "$CONTROL/scripts/query-technology-intelligence.sh" --query "sentence embeddings"
  8. Package-registry file TI (npm/PyPI-shaped fixtures — no registry network in CI): use Skill package-registry-ti or:

    COMPASS_TI_PROVIDER=package-registry-file \
      "$CONTROL/scripts/query-technology-intelligence.sh" --query "schema validation"
  9. Live package-registry TI (Captain local — npm + PyPI):

    COMPASS_TI_PROVIDER=package-registry \
      "$CONTROL/scripts/query-technology-intelligence.sh" --query "schema validation"
  10. Batch categorized Stars (offline ML from manual labels — no live training in CI):
"$CONTROL/scripts/categorize-github-stars.sh" --source fixtures   # offline test path
# or after refresh:
"$CONTROL/scripts/categorize-github-stars.sh" --source ti-cache
COMPASS_TI_PROVIDER=github-stars-categorized \
  "$CONTROL/scripts/query-technology-intelligence.sh" --query "react forms"
  1. Skill learning loop (M19/M24 — staging + harness + drafts / improvement proposals):
"$CONTROL/scripts/northstar" skills learn \
  --repo "$SANDBOX" \
  --source fixtures \
  --objective "accessible react forms"
# equivalent:
"$CONTROL/scripts/run-skill-learning-loop.sh" \
  --repo-root "$SANDBOX" \
  --source fixtures \
  --objective "accessible react forms"

Review drafts under .agent/capabilities/candidates/skill-drafts/ and improvement proposals under skill-improvement-proposals/. Live install still requires Captain approval (skill-learning-loop Skill).

  1. Live Hugging Face Hub TI (Captain local — Hub models API):
COMPASS_TI_PROVIDER=huggingface-hub \
  "$CONTROL/scripts/query-technology-intelligence.sh" --query "sentence embeddings"
# optional auth for gated models:
# COMPASS_HF_HUB_TOKEN=hf_... COMPASS_TI_PROVIDER=huggingface-hub ...

Output

  • JSON list of normalized CandidateCapability payloads
  • Plan section Technology Intelligence Candidates when planning with COMPASS_TI_PROVIDER=github-stars, github-stars-cached, github-stars-categorized, huggingface-file, huggingface-hub, package-registry-file, or package-registry

Prohibited actions

  • Running live TI in CI or as default provider
  • Setting approved_for_execution: true
  • Auto-cloning, installing, or executing external repositories
  • Live Hugging Face Hub or npm/PyPI registry network calls from CI defaults
  • Assuming TI/learning scripts exist inside product/sandbox checkouts

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.