hassan-mohiddin/freeflow / .deprecated/skill-tooling-pre-rewrite/skills/write-skill
write-skill
VerifiedUse when creating or revising an agent skill.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.7 KB |
| references/activation-boundaries.md | markdown | 2.9 KB |
| references/agent-first-instructions.md | markdown | 3.3 KB |
| references/development-loop.md | markdown | 2.2 KB |
| references/progressive-disclosure.md | markdown | 1.9 KB |
| scripts/lib/skill-author-core.mjs | script | 6.2 KB |
| scripts/skill-author.mjs | script | 2.0 KB |
references/activation-boundaries.md
View on GitHub# Activation Boundaries
A description should make the skill available at the earliest useful moment without requiring the agent to know language introduced by the skill itself.
## Decide The First Read
Before writing the description, identify:
- the earliest natural prompt or state where the method can change behavior;
- what the agent can observe at that moment;
- what context is guaranteed to exist before activation;
- the cost of missing that first read;
- the cost if a nearby prompt also loads the skill.
Do not optimize for repeated reads. The description’s main job is to make the method available before the first decision it should influence.
## Use Observable Cues
Include:
- the skill’s job;
- verbs and phrases users naturally use for that job;
- concrete situations visible in the prompt or immediate context;
- distinguishing cues when neighboring skills are similar.
Avoid:
- vocabulary defined only in the body or an optional sibling skill;
- future-duration guesses such as “sustained”;
- unexplained classifications such as “consequential” or named workflow phases;
- generic “help with” or “improve” language;
- marketing claims and exhaustive trigger taxonomies.
If guaranteed base context defines a term, the description may use it. Otherwise prefer plain observable language.
## Prefer Recall Without Hijacking
A broad description is useful when the body can quickly recognize a nearby case and return control without adding ceremony or changing the task.
Test two kinds of nearby behavior:
- **Non-trigger:** the skill should clearly not load.
- **Safe read:** loading is acceptable, but the body must answer, defer, or exit without applying the wrong method.
Do not narrow a description solely to prevent harmless reads. Do not broaden it so far that the skill routinely overrides the actual owning job.
## Boundary Examples
Broad core trigger:
> Use when committing work.
Over-specified trigger that hides the same job behind workflow details:
> Use after final verification when a planned checkpoint requires a locally reviewable commit.
Broad safe-read trigger:
> Use when working with software design—especially boundaries, interfaces, ownership, state, and failure behavior.
The body can exit for a local change whose interface remains sound. The description does not need to encode every structural-pressure case.
## Evidence
Separate delivery from behavior:
- native activation proves the host read the exact skill snapshot from a natural prompt;
- explicit body delivery tests whether first-read instructions are understandable;
- composition tests declared dependencies;
- multi-turn evidence tests retained use after activation;
- artifact or transcript evidence shows whether activation helped or hijacked behavior.
Before a readiness claim, preserve the earliest useful positive prompt and at least one close nearby case. The nearby case may pass through non-activation or safe behavior, depending on the intended boundary.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |