agentleFS
thituongh699-hashthituongh699-hash/oriward/.agents/skills/career-discovery/SKILL.md

career-discovery

Skill2 starsChanged 7 days agoMITView on GitHub

Recover work, business, education, projects and informal responsibilities through natural conversation, preserving user reports separately from inferred directions. Use for experience discovery and translation, not vacancy search.

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:

    Read context brief, current topic, intake next and experience index.

    line 10
  2. Distinctive:

    Follow AGENTS.md and the decision contract for recommendations, not ordinary experience organisation.

    line 23

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Career discovery

Deep discovery is the first-use default, from a story, resume or authorized history. Only an existing resume plus an explicit request for immediate search and application permits the direct route; offer to deepen the library too. Never invent titles, ownership, dates, tools, metrics or outcomes.

  1. Read context brief, current topic, intake next and experience index; page the full index, then load relevant details. Load experience and history contract when starting discovery or resuming history intake. Reuse known answers. Use $career-mapping when directions are wanted.
  2. Proactively offer to build an initial timeline from specified past chats. Reuse existing scoped authorization without asking again. Read authorized user messages before asking questions they already answer. The CLI supports exact export files; host chat tools require explicit chat scope. No available history access, deferral or refusal means start from the user's story, not a blocked intake. Never scan all private chats or treat assistant text as personal achievements.
  3. Actively notice vague duties, unexpected responsibility, difficult decisions, workarounds, setbacks, repeated tasks and informal contributions the user dismisses. Ask one key question per turn, not several questions disguised as one. Choose a concrete episode or personal contribution before cosmetic details. Reflect known facts briefly, then follow the user's answer.
  4. Develop an episode across turns: purpose, personal actions/decisions, obstacles/responses, known outcomes or lessons. These are recall lenses, not mandatory fields. Distinguish personal work, team output and AI assistance. Missing numbers can remain qualitative; hesitation/refusal means pause that lead. Do not suggest an invented accomplishment for agreement.
  5. Save clear self-reports before the next question through experience add/revise, using the same stable event ID as details emerge. No attachment, exact KPI or approval is required. Preserve original account, provenance and versions; model guesses remain hypotheses. Corrections update the event; role-specific wording belongs in resume versions. No per-phrase approval is needed.
  6. Persist valuable unfinished leads with intake clue: one question, reason, source locator and optional experience ID. Priority 1: contribution/conflict; 2: concrete action/decision/result; 3: another stage. Consolidate useful existing open_questions; do not create a question for every empty field. After saving an answer, mark its clue answered; user pauses are later, refusals declined. Reopen only when the user returns to it. Save topic/next action with onboarding topic.
  7. Resume with intake next; page intake clues to avoid repeats. Current user topic overrides ranking. Inspect current experience before asking; retire already-answered or stale questions. Do not automatically reopen deferred leads or excluded experiences. The queue is a conversation aid, never a verification or drafting gate.
  8. After a useful episode or pause, offer the organised account and another relevant stage or stopping. An empty queue does not prove a complete history. Naturally check the education/career skeleton without assuming every category applies. Deliver a full draft when asked, even with unresolved details; report saved IDs/versions briefly.

Career Twin Core contract

  • Save capability/direction hypotheses with belief propose, source_type:model_inference and actual experience links. Grounded factual wording needs no new confirmation stage.
  • Preserve meaningful uncertainty without downgrading clear self-reports. belief confirm records direct user agreement only, never a drafting prerequisite. Legacy approved_for_resume=false is not rejection.
  • Follow AGENTS.md and the decision contract for recommendations, not ordinary experience organisation.

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.