Frames coding-agent work sessions with explicit intent capture and drift monitoring. Use when a session transitions from planning/Q&A to implementation for c...
Security Analysis
high confidenceThis is an instruction-only skill whose instructions, requirements, and behavior are consistent with its stated purpose (framing coding sessions and monitoring scope drift); it requests no credentials and has no install spec, but it does describe writing structured blocks into session transcripts which you should verify are stored/processed according to your privacy needs.
Name/description match the SKILL.md: the skill's sole function is to capture intent frames, monitor drift, and emit structured intent/resolution blocks. It does not require unrelated binaries, environment variables, or credentials.
Instructions stay within the stated scope (capture/monitor/resolve intent for coding work). They do instruct the agent to emit structured blocks into the session transcript and to detect the Entire CLI if available; this means intent blocks may be recorded in transcripts and later mined by tooling (learning-aggregator). That is expected for the skill's purpose but is a privacy/retention consideration.
No install spec or code files are present; the SKILL.md contains example commands (npx skills add ...) for adding related packages but those are guidance only. There is no automatic download/install described in the skill itself.
The skill declares no required environment variables, no credentials, and no special config paths. Nothing requested is disproportionate to its described functionality.
The skill is not configured always:true and does not request permanent presence or elevated privileges. It does describe interoperability with local tooling (Entire CLI and learning-aggregator) but does not instruct modifying other skills or system-wide settings.
Guidance
This skill appears coherent and low-risk: it only emits structured intent/check/resolution text and asks you to confirm before proceeding. Before installing/use, verify provenance (source/homepage is missing here) and double-check where session transcripts and checkpoint branches are stored and who can access them—if you run Entire CLI or learning-aggregator, confirm whether those tools upload transcripts off-host or make them available to others. If you have sensitive data that might appear in intent frames, avoid including it (or confirm retention/processing policies). Also review any external packages (the SKILL.md references a pskoett package) before running installation commands.
Latest Release
v1.0.0
- Updated installation instructions to reflect the correct subpath: use `pskoett-ai-skills/skills/intent-framed-agent`. - Expanded the Entire CLI integration section with details on how intent frames and drift events are mined for learning signals. - Added a section explaining how intent frames are used as structured learning signals with tools like `learning-aggregator --deep`. - Revised the interoperability section to clarify concurrent use with the `context-surfing` skill, providing a detailed description of their complementary roles and precedence rules. - Clarified what artifacts and outputs this skill produces and how they interact with related skills.
More by @pskoett
Published by @pskoett on ClawHub
