Skillget

behavior-enhancer-generic

agent
Security Audit
Fail
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Fail
  • rm -rf — Recursive force deletion command in adapters/claude-code/hooks/register.test.ts
  • rm -rf — Recursive force deletion command in adapters/claude-code/hooks/register.ts
  • process.env — Environment variable access in adapters/codex/hooks/hooks.json
  • fs.rmSync — Destructive file system operation in adapters/codex/hooks/post_tool_use.cjs
  • fs module — File system access in adapters/codex/hooks/post_tool_use.cjs
  • exec() — Shell command execution in adapters/codex/hooks/pre_tool_use.cjs
  • fs module — File system access in adapters/codex/hooks/pre_tool_use.cjs
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Behavior discipline for coding agents: read-before-write enforcement, failure convergence, write verification with auto-rollback, and a high-risk command gate. Ships FULL plugin/hook adapters for DSH, Claude Code, Codex, and Pi, plus a host-agnostic SOFT skill with ADAPTIVE adapter guidance.

README.md

behavior-enhancer-generic

English · 中文

Listed on ClaudePluginHub

The behaviour design of dsh-behavior-enhancer, shipped as a cross-host dual-form product:

  • Plugin form (FULL): where the host exposes runtime interception, you get real enforcement — vetoes, write verification, auto-rollback;
  • Skill form (SOFT): where the host only gives an agent-facing channel, you get behavioural guidance (self-discipline rules).

Same semantics, different enforcement strength. No FULL claims we cannot back, and no pretending SOFT is FULL.

Which host gets which form

Host Form What to install Status
Claude Code FULL (8 features) adapters/claude-code/ — plugin (F3b is implemented as a concurrency-window veto) engine-level tests 14/14, dogfooding
Codex FULL (8 features) adapters/codex/ — config.toml hooks or plugin manifest protocol tests 13/13, ignition-tested on the desktop build
Pi FULL (7 features + half of F9) adapters/pi/ — pi extension protocol tests 11/11, live-tested on a real install
DSH FULL adapters/dsh/ — installable plugin (all modules of the original, plus F1 read-before-write and two fixes) all-module mock-ctx tests pass, real-install test pending
Any other host SOFT + ADAPTIVE skill/SKILL.md §8: the agent self-builds an adapter from core/adapter-template.md (conservative testing, conservative FULL, written report) published form

Install on Claude Code (one command):

/plugin marketplace add Zoria-Lind/behavior-enhancer-generic
/plugin install behavior-enhancer@zoria-behavior

For local development you can instead use claude --plugin-dir "<repo>\adapters\claude-code" or
CLAUDE_CODE_PLUGIN_DIRS=...\adapters\claude-code (see
adapters/claude-code/README.md) — pick one, never both.

The four-state activation model

Every feature activates independently, and its state is decided by the host's capability surface — never by user or agent preference:

State Meaning
NATIVE The host already does this (only claimed when measured); the skill stays quiet
FULL Plugin-level takeover (veto / write verification / rollback)
SOFT Behavioural guidance (instructions, flow, decision trees)
DISABLED We do not pretend a capability exists

There are no if DSH / elif ClaudeCode branches — only capabilities. Full design in
generic-skill_plan_v1.md; the feature × capability matrix in
core/audit.md; cross-host shared semantics (tokens, pattern tables, thresholds, TTL, tier ladder) in
core/semantics.md.

Note: those design documents (core/, generic-skill_plan_v1.md) are Chinese-only for now. English readers can work from this README plus the per-adapter READMEs.

Relationship to the DSH plugin

The DSH adapter started from the original plugin dsh-behavior-enhancer (a separate repo) and then added the generic work. From there the two evolve independently — there is no sync mechanism: when one side produces a better rule, the other copies it by hand if it is worth it. Install one or the other, never both
(see adapters/dsh/README.md).

Repository layout

behavior-enhancer-generic/
├── .claude-plugin/  # Claude Code marketplace manifest (marketplace.json, name: zoria-behavior)
├── core/            # behaviour semantics + capability matrix (four states) + generic adapter template (ADAPTIVE mode)
├── skill/           # SOFT artifact: SKILL.md + per-host resident fragments (claude-code / codex / pi / cursor)
├── adapters/        # FULL artifacts: claude-code / codex / pi (DSH is only a pointer back to the original repo)
├── publish/         # Agensi publishing artifacts (the SOFT skill package)
└── README.md / README.zh-CN.md   # English (this file) / Chinese

Tests

cd adapters/claude-code && claude plugin validate . && claude plugin test .   # 14 engine-level tests
cd adapters/codex && node test/adapter.test.cjs                              # 13 protocol tests
cd adapters/pi && node test/adapter.test.cjs                                 # 11 protocol tests

Links

License

MIT — see LICENSE. The rules started from the DSH version and have evolved independently since.

Reviews (0)

No results found