Skill Detail
thought-chain-planner
Planning multi-step reasoning and execution flows that help agents move from vague intent to concrete action.
What problem it solves
An agent can have good local reasoning and still fail if it cannot structure a longer chain of thought and action. thought-chain-planner solves the problem of unstructured reasoning by designing multi-step plans that connect intent, analysis, and execution.
How it works
- •Start with the desired end state rather than isolated next steps.
- •Lay out the reasoning chain needed to get from intent to action.
- •Separate planning from execution enough to reduce thrashing.
- •Use intermediate checkpoints to validate whether the chain still makes sense.
- •Adjust the plan when new evidence or constraints appear.
Use case from logs
Planning a delegated refactor before execution
Context: When a change carries technical uncertainty, jumping directly into implementation increases thrash and weakens delegation quality.
What happened: CAS first framed the expected outcome, selected the right execution path, and delegated the refactor through the task layer instead of collapsing reasoning and execution into one opaque move.
Outcome: That planning chain made the later evaluation meaningful because predictions, delegation, and execution were linked from the beginning.
Source
CAS + OpenCode: Conscious Coding Agents — content/blog/cas-opencode-conscious-coding-agents.md
CAS predicts: “This refactor breaks 3-5 tests” ... decides: Use Build agent ... delegates via Task tool → OpenCode.
GitHub
Code examples for this skill will link out to GitHub once the public repo is ready.