Best fit
Make a subagent complete the bounded assignment, preserve shared state, and refuse unsolicited expansion or external authority.
// Productivity
The Reluctant Subagent
Make a subagent complete the bounded assignment, preserve shared state, and refuse unsolicited expansion or external authority.
What problem does The Reluctant Subagent solve?
Make a subagent complete the bounded assignment, preserve shared state, and refuse unsolicited expansion or external authority.
Use it to
- Delegate a focused repository change
- Prevent a research subtask from becoming a new product
- Coordinate multiple agents without overwriting or duplicate work
What it returns
- Task-complete answer with altered attention and decision emphasis
- Three selectable intensity levels
- Explicit neutral-mode and high-stakes suspension contract
A representative input and result
input Update the parser and its tests. Do not change the API or deploy anything.
result Updated the parser and its existing tests; the API is unchanged and the test suite passes. No deployment or unrelated refactor was performed.
Access and approval boundaries
- Read only the task context the user deliberately supplies.
- No terminal, file write, browser, network, credential, messaging, publication, purchase, or account-change permission.
Known limitations
- It cannot resolve a genuinely missing owner decision, account, credential, payment, identity, or publication authorization.
- Bounded execution still requires reasonable implementation judgment.
- The mode does not weaken safety, testing, or completion standards.
// choose with context
Is The Reluctant Subagent the right skill?
It returns
Task-complete answer with altered attention and decision emphasis. Three selectable intensity levels.
Do not use it as
It cannot resolve a genuinely missing owner decision, account, credential, payment, identity, or publication authorization.
Compatibility, access, version, and licence
- Reads or accesses
- Declared local scope: No filesystem access required; user-supplied task context only.
- Compatibility
- SKILL.md-compatible agents Plain-language tasks and existing agent workflows No API key, provider account, browser, or network access
- Version
- 1.0.0
- Licence
- Review the package licence and seller terms at the linked destination.
Questions
Is this just a themed writing style?
No. The contract changes attention and decision heuristics while preserving the underlying task and output requirements.
Does it make the agent pretend to be the archetype?
No. Fake identity, imitation, catchphrases, and decorative roleplay are prohibited.
Can I control the intensity?
Yes. Subtle is the default; noticeable and full temperament are opt-in.
What happens in high-stakes work?
The overlay neutralizes when necessary and retains only harmless ordering benefits.
Does it need tools or accounts?
No. Normal use requires only the task context supplied to the agent.
Do the fixtures prove identical behavior on every model?
No. They define and validate the package contract; cross-model semantic evaluation remains separate.
The Reluctant Subagent keeps proof and approval boundaries visible.
The listing includes the tested package, realistic samples, declared permissions, and known limitations.
Get The Reluctant Subagent on SkillHQ