dharma-autonomous-build

Executes bounded overnight autonomous cleanup and build lanes with explicit stop conditions.

1|3|Updated Mar 9, 2026
One-click install
npx skills add https://github.com/AIKAGRYA/dharma_swarm --skill dharma-autonomous-build-aikagrya
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: dharma-autonomous-build
Source: https://github.com/AIKAGRYA/dharma_swarm/tree/main/mode_pack/claude/autonomous-build
Command: npx skills add https://github.com/AIKAGRYA/dharma_swarm --skill dharma-autonomous-build-aikagrya

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Long-running autonomous coding sessions tend to drift out of scope, touch protected hot paths, or silently widen into adjacent domains. This Skill keeps an agent executing a pre-defined cleanup or build lane overnight while enforcing lane boundaries, stop conditions, and a truthful control trail after every tranche. ## Core Features & Use Cases - Bounded autonomous execution: Keeps iterating on an already-decomposed lane without reopening scope each cycle, stopping on ambiguity instead of widening. - Hot-path protection: Explicitly forbids widening from non-TUI cleanup into TUI or dashboard hot paths and treats untracked files as not merge-safe. - Structured control trail: Requires a six-part report after each tranche covering current lane, tranche completed, files changed, validation performed, residual risks, and the next bounded seam. - Use Case: An operator defines a cleanup lane in a control document before leaving for the night; the agent works through tranches, updates the control docs and frontmatter as it goes, and stops with a recorded note when it hits unresolved ownership or path-coupling ambiguity. ## Quick Start Activate the dharma-autonomous-build mode and execute the next bounded tranche of the cleanup lane defined in the overnight control document, then report files changed and residual risks.

Frequently Asked Questions about dharma-autonomous-build

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an autonomous overnight coding session safely?▼

Define the lane, hot-path exclusions, and stop conditions in a control document first, then activate this mode. The agent executes bounded tranches, records files changed and validation performed, and stops on ambiguity instead of widening scope.

How do I keep an AI agent from drifting out of scope during long tasks?▼

Use explicit lane boundaries and stop rules that forbid widening into protected hot paths such as TUI or dashboard code. The agent must stop and record unresolved ownership, path-coupling, or authority ambiguity rather than guessing.

What should an autonomous agent report after each work tranche?▼

Each tranche report contains six items: current lane, tranche completed, files changed, validation performed, residual risks, and the next bounded seam. This creates a truthful control trail for operator review.

When should I not use autonomous build mode?▼

Do not use it when the lane is not yet decomposed, stop conditions are unwritten, or the work touches protected hot paths like TUI or dashboard code. Resolve ownership and scope questions interactively first.

Does autonomous mode treat existing files as merge-safe?▼

No. The rules explicitly state that a file existing does not mean it is tracked and merge-safe. The agent must verify tracking status and update control docs and frontmatter as the lane evolves.