Plugin Claude Code
Plugin marketplace listing 1 plugin: samsara.
Plugin Claude Code
Plugin marketplace listing 1 plugin: samsara.
Plugin Claude Code
向死而驗 — Death-first development workflow. Existential accountability for every line of code.
Instructions file CodexOpenCode
AGENTS.md instructions for roymondliao/samsara, covering project conversation guidelines, python, infrastructure and when a command fails or produces unexpected output.
Instructions file
Claude Code instructions for roymondliao/samsara: MUST read and STRICTLY follow AGENTS.md — these are mandatory execution conventions, not suggestions.
Agent
Staff-level workflow decision authority for Samsara auto mode — answers human-equivalent gate questions and is the sole writer of validated append-only decisions.
Agent
Structural code quality review agent — assesses structural truth-telling against 9 yin principles (S/O/L/I/D + Cohesion/Coupling/DRY/Pattern). Produces PASS/PASSWITHCONCERNS/FAIL/UNKNOWN verdicts with mandatory file:line evidence on any non-UNKNOWN output.
Agent
Yin-side code review agent — asks deletion before correctness, identifies dishonest naming and silent rot paths. Determines execution model and loads domain-specific reference files. Returns UNKNOWN for unrecognized or unsupported domains.
Agent
Death-test-first implementer — writes death tests before unit tests, produces scar reports, enforces STEP 0 prerequisite questions.
Agent
Explores committed build, configuration, data-flow, and external-system evidence.
Agent
Explores committed codebase module boundaries, nodes, relationships, and public interfaces.
Agent
Analyzes committed code for silent failure, hidden coupling, assumptions, and death impact after structural evidence exists.
Hook
Runs when a session starts on startup, clear and compact, running an inline shell check (2 commands). From roymondliao/samsara.
Skill Claude CodeCodex
Use when entering a Git project for the first time or when its committed HEAD no longer matches the recorded codebase map.
Skill Claude CodeCodex
Use when existing deployed or runtime behavior produces an observed failure, alert, or user-reported regression.
Skill Claude CodeCodex
Use when state-changing work has a proven bounded damage radius, no unresolved design question, and deterministic verification.
Skill Claude CodeCodex
Use when a plan with tasks exists and you need to execute implementation — requires index.yaml and tasks/ directory.
Skill Claude CodeCodex
Use when implementation is committed and task scar reports may contain unresolved, accepted, deferred, blocked, or cross-task wounds before shipping.
Skill Claude CodeCodex
Use when a decision-maker needs comparable engineering options, broader scope, or an independent view before making a difficult technical judgment.
Skill Claude CodeCodex
Use when approved pre-thinking output must become an implementation task graph and acceptance contract without reopening design decisions.
Skill Claude CodeCodex
Use when research output is complete and planning would otherwise need to assume design, task-shape, structural boundaries, or evaluation details before implementation planning.
Skill Claude CodeCodex
Use when starting new feature work, investigating a problem, or when the user describes something they want to build — before any planning or implementation.
Skill Claude CodeCodex
Injected at session start — establishes samsara axiom, agent constraints, and skill discovery for the session.
Skill Claude CodeCodex
Use when Iteration has committed a ready-for-validation checkpoint and one candidate must be validated, documented, and prepared for delivery.
Skill Claude CodeCodex
Use when creating, revising, or reviewing Samsara skills.