Managing littlepowers
Proportional planning, recovery, and engineering-discipline protocol for Codex, Claude Code, Qoder, and OpenCode.
npx -y skills add clsaa/littlepowers --skill managing-littlepowersAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 22 days oldThe repository was created 22 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Inspect or manage Littlepowers ledger status, health, pause, resume, cancel, replacement, cleanup, or troubleshooting. Do not edit product code.
SKILL.md
4.7 KB, ~1.0k tokens by cl100k_base, as published. Nobody here has run it
Managing Littlepowers
Operate the recovery ledger without assuming the user's project contains the Littlepowers source tree.
Resolve <state-cli> as follows:
- Claude Code expands
${CLAUDE_PLUGIN_ROOT}/scripts/littlepowers_state.py. - Qoder CLI expands
${QODER_PLUGIN_ROOT}/scripts/littlepowers_state.py. - In Codex or OpenCode, resolve
../../scripts/littlepowers_state.pyfrom this loadedSKILL.mdpath.
If that path disappeared after a plugin replacement, do not continue from memory. Use the host-native installed-plugin listing described by using-littlepowers to resolve exactly one active Littlepowers root, reread this skill there, and then run the state CLI. If resolution is missing or ambiguous, stop and use a new task or session after installation.
Use an available Python 3 launcher. Run doctor for trust, ignore, schema, and validity checks; run show --json for status. These commands are read-only.
Report the ledger's progress as written. If it is absent, describe the current task, completed checkpoints, and next action without inventing a percentage. Also report the stored contract, coverage, baseline, and fidelity summary. These are last-known facts; status and hook commands do not refresh files.
If the lock is reconcile_required, explain that active legacy implementation
cannot continue until the approved contract is bound and its Plan Map passes.
If it is drifted, use check-contract for an explicit observation and require
approved rebind for changed content. Do not repair either condition by editing
.littlepowers/state.json.
Pause, resume, handoff, cancel, checkpoint, and complete require the workflow ID and current revision from show --json. Pass both with --workflow <id> --expect-revision <revision>. Use the newly returned revision after every mutation.
Completion is valid only from phase=verify after a current bound contract,
passing coverage, a recorded Verification Record, passing work-unit and
approved-outcome fidelity verdicts, required code-quality approval, and zero
blockers. The completion command performs fresh deterministic checks and lists
all failures.
Examples:
<python> <state-cli> doctor
<python> <state-cli> show --json
<python> <state-cli> check-contract \
--workflow <id> --expect-revision <revision>
<python> <state-cli> bind-contract \
--workflow <id> --expect-revision <revision> \
--artifact <contract.md> --approval-kind review-gate
<python> <state-cli> validate-plan \
--workflow <id> --expect-revision <revision> --artifact <plan-or-shape.md>
<python> <state-cli> record-verification \
--workflow <id> --expect-revision <revision> --artifact <verification.md>
<python> <state-cli> pause --workflow <id> --expect-revision <revision>
<python> <state-cli> resume --workflow <id> --expect-revision <revision>
<python> <state-cli> handoff --workflow <id> --expect-revision <revision> \
--target-root <absolute-path> \
--target-workflow <target-id> --target-revision <target-revision>
<python> <state-cli> cancel --workflow <id> --expect-revision <revision>
<python> <state-cli> start --replace \
--workflow <id> --expect-revision <revision> \
--objective "<replacement outcome>" \
--phase <brainstorm|shape|execute> \
--next-action "<next observable action>"
Use handoff only for an actual transfer to a different workspace root. The target ledger must already exist and be active. The command verifies that explicit target, then cancels only the source workflow and records a pointer; it never writes the target or changes the current task root. Continue in a new task or session rooted at the target and verify its current state there. Do not scan sibling worktrees to discover a destination.
On a conflict, reload and explain which workflow or revision changed. Do not retry blindly.
Starting with --replace archives any current ledger, including a terminal one, and requires its workflow ID and revision. Use it for an active or paused ledger only when the latest request clearly switches objectives; use it for a terminal ledger when beginning the next tracked objective. Uninstalling the plugin does not remove .littlepowers; preserve, archive, or delete that local directory only when the user requests cleanup and confirms the exact workspace.
Schema-3's first successful mutation of a legacy ledger writes one matching
pre-schema3 archive. A 1.1 runtime cannot read schema 3; preserve the current
schema-3 state separately and restore the exact matching archive only during an
explicit, stopped-writer rollback.
What ships with it: 1 file
204 B alongside SKILL.md
agents/
- openai.yaml204 B