two tiers
The catalog is framework-free. The kit operates Suspec concepts. Keep repo policy local.
suspec-skills
Two tiers of task-scoped guides: a framework-free catalog and a Suspec-coupled kit.
The catalog (suspec-skills) installs into any repo via npx skills. The kit ships in suspec-starter-kit. Load only the guide in scope.
- 01Source
- 02Install
- 03Load
- 04Run
index
guide files
catalog
11
kit
13
file
SKILL.md
Guidance only. Read the file before it shapes a team workflow.
skills add
$ npx skills add jcosta33/suspec-skills
This installs the framework-free catalog into any repo. The Suspec kit ships separately in suspec-starter-kit. Keep repo-specific commands in AGENTS.md.
catalog · review stances
Review stances
Framework-free catalog stances for judgment, evidence, security, and debugging. Installed with npx skills.
markdown onlyon demand
suspec-skills catalog6 guides
adversarial-review
review another agent's work refute-by-default
persona-challenger
pressure-test a proposal before build work starts
persona-surveyor
survey patterns across examples
empirical-proof
back completion claims with pasted output
security-review
audit a change for security exposure
debugging
find a defect's root cause from runtime evidence
catalog · working disciplines
Working disciplines
The catalog's everyday methods: map a codebase, plan a change, ship a PR, keep output terse, stabilize a flaky test. Also framework-free.
kit · suspec-coupled
The Suspec kit
These skills operate Suspec concepts — task packets, review packets, specs, findings — so they are not framework-free. They ship in suspec-starter-kit under .agents/skills/, not in the npx skills catalog. The write-* family is opt-in task-implementation depth.
suspec-starter-kit/.agents/skills13 guides
implement-task
implement a Suspec task packet
review-output
build the review packet for a finished task
spec-check
check a spec against the core checks
split-work
split a spec into non-colliding task packets
save-findings
save what a task taught as findings
write-feature
add new behavior
write-fix
fix a reproduced defect
write-refactor
restructure without changing behavior
write-rewrite
rewrite with a recorded behavior change
write-migration
move from one API to another
write-performance
improve a measured bottleneck
write-testing
add focused tests
write-documentation
write human-facing docs
authoring.guide — write your own
How to write a Suspec skill
Use a clear description, a self-contained body, visible output, and the AGENTS.md command contract.
security.note — no runtime
Why there is no runtime
Review the guide before installing and pin to a commit when you use it in a team repo. For scaffolding and checks, use suspec-cli.