Charlie
Community@mistakenot · London
Charlie (mistakenot) builds agent-orchestration skills for task planning, multi-model code review, git release management, and engineering knowledge mining.
Agent Skills by Charlie
Showing 47 vetted skills indexed across 2 GitHub repositories.
mistakenotborrow-from-oss
Mines watched open-source repositories for changes and ranks borrowable ideas in a YAML backlog.
mistakenotrich-doc
Creates single-file interactive HTML documents with tabs, mermaid diagrams, and review threads.
mistakenotrequest-claude-review
Runs Claude Code reviews of task planning docs and resolves review comments.
mistakenotnew-task-quick
Plans a task end-to-end in one unattended pass, producing plan.html and context.md with a Codex review.
mistakenotnew-epic
Creates an epic.html planning document with direction, guard rails, and a sequenced task breakdown.
mistakenotnew-task
Creates a plan.html requirements document for a new software task.
mistakenotcode-review
Reviews code changes with severity-labeled findings across correctness, security, performance, and testing.
mistakenotdelegate-task
Dispatches planned tasks to background coding-agent workers in isolated git worktrees via herdr.
mistakenotrequest-codex-review
Runs a Codex code review of task planning docs and resolves the resulting comments.
mistakenotdelegate
Launches background coding agents in isolated git worktrees via herdr with verified permission flags.
mistakenotresolve-comments
Resolves inline review comment threads in markdown and HTML planning documents.
mistakenotexecute-task
Implements planned tasks autonomously using coordinator-subagent phases, worktrees, and pull requests.
mistakenotcommit-task
Verifies planning document completeness and commits them to the main branch.
mistakenottask-feedback-analyser
Extracts recurring patterns from task feedback into generalizable workflow rules.
mistakenotcomplete-task
Finalizes a feature branch and merges its pull request to main.
mistakenotaddress-feedback
Resolves open GitHub PR review threads by fixing code, replying, and resolving threads.
mistakenotrequest-council-review
Runs parallel read-only reviews of planning docs by Claude, Codex, and Grok, then merges and resolves findings.
mistakenotstatus-report
Monitors herdr background coding agents, health-checks their launch flags, and reaps finished workers.
mistakenotnew-plan
Writes the Plan tab in plan.html and backfills acceptance criteria traceability for a task.
mistakenotrequest-grok-review
Runs Grok CLI code reviews on task planning docs and resolves resulting comments.
mistakenotnew-solution
Generates context, verification, and solution design tabs for approved task requirements.
mistakenotreview-task
Reviews task planning documents and inserts structured inline comments flagging problems against the codebase.
mistakenoteval-engineer
Builds and validates A/B evaluation harnesses that compare skill versions on cost and quality.
mistakenotdomain-modelling
Builds and maintains a DDD ubiquitous language glossary with validation and ER diagram generation.
Frequently Asked Questions About Charlie
FAQPage SchemaWhat tasks can I accomplish with mistakenot's skills?▼
You can plan tasks and epics with structured HTML docs, execute them via background workers, run multi-model code reviews, tag semantic releases on GitHub, write postmortems, mine session history for feature ideas, and extract customer demand from call transcripts.
Who are these skills designed for?▼
Software engineers and technical leads running agent-assisted development who need disciplined task planning, independent review from multiple coding models, release versioning, and systematic capture of lessons learned across sessions and completed work.
How do the planning and execution skills fit together in practice?▼
Start with new-task or new-task-quick to create requirements, solution, and plan docs, optionally request Claude, Codex, or council review, commit the docs, then delegate-task dispatches execute-task to a background worker that implements phases and opens a PR.
What prerequisites or dependencies do these skills require?▼
Skills assume a git repository with GitHub access for releases and PRs, background worker support via herdr for delegation, and access to coding models such as Claude, Codex, and Grok for reviews. The pd-test skill requires Chrome via agent-browser.
How do the reflection and learning skills improve future work?▼
playbook-observe mines completed task runs into observations, playbook-refine distils them into evergreen rules, and playbook-search retrieves relevant rules at planning time. learning-diary separately captures techniques and breakthroughs from git history and PRs.