leos-agent 6.1.0 → 6.1.1

package/README.md

@@ -0,0 +1,43 @@
+<!-- Generated by scripts/render_adapters.py; do not edit. -->
+
+# Leo's Agent
+
+Leo's Agent is a portable agent operating policy: cost-tiered model routing, specialist subagent roles, process skills, execute-then-review discipline, and a narrow catastrophic-command guard.
+
+This npm package is the **OpenCode** distribution. Claude Code, Codex, Cursor, and Hermes each install it through their own plugin system — see [the repository](https://github.com/foxhatleo/leos-agent) for those.
+
+## Install
+
+Add it to `~/.config/opencode/opencode.json`:
+
+```json
+{ "$schema": "https://opencode.ai/config.json", "plugin": ["leos-agent"] }
+```
+
+Start a new OpenCode session. The plugin registers the skills directory, the 6 subagent roles, and the operating policy, and installs the bash deletion tripwire.
+
+If the skills do not appear, register the path by hand — npm installs into OpenCode's own cache and the location varies by version:
+
+```json
+{ "skills": { "paths": ["~/.cache/opencode/node_modules/leos-agent/skills"] } }
+```
+
+## Model tiers
+
+Tier names describe the kind of work, not a fixed provider model.
+
+| Tier | Model | Effort |
+|---|---|---|
+| Fable | `moonshotai/kimi-k3` | native default |
+| Opus | `moonshotai/kimi-k3` | native default |
+| Sonnet | `z-ai/glm-5.2` | native default |
+| Haiku | `z-ai/glm-5.2` | native default |
+
+Fable and Opus collapse onto one model here, so `expert` is not registered as an agent and escalation caps at Opus. Retier by editing `config/models.json` and re-running `scripts/render_adapters.py`.
+
+## Links
+
+- [Repository and full documentation](https://github.com/foxhatleo/leos-agent)
+- [Operating policy](https://github.com/foxhatleo/leos-agent/blob/main/plugins/leo/skills/using-leo/SKILL.md)
+
+MIT licensed.

package/adapters/opencode/plugin.js

@@ -1,4 +1,6 @@
-// Leo's OpenCode bridge: registers the leo:* skills dir, injects the
+// Leo's OpenCode bridge: registers the skills dir by path (OpenCode names
+// each skill from its own frontmatter, so there is no leo: prefix here as
+// there is on Claude Code and Hermes), injects the
 // generated subagent roster from adapters/opencode/agents.json, assembles
 // the using-leo policy plus the OpenCode mapping appendix and hands it to
 // OpenCode through config.instructions (belt) and the

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "leos-agent",
-  "version": "6.1.0",
+  "version": "6.1.1",
   "description": "Leo's agent operating policy: cost-tiered routing, subagent roles, review gates, guardrails.",
   "type": "module",
   "main": "adapters/opencode/plugin.js",

package/scripts/render_adapters.py

@@ -252,7 +252,11 @@ def _mapping_docs(config):
         "coarse `rm -rf` bash denies as a stopgap for opencode#5894 (unconfirmed whether "
         "`tool.execute.before` also intercepts subagent bash); the precise tripwire stays "
         "`hooks/bash-guard.py` on the primary agent.\n"
-        "\nNo `EnterWorktree` tool exists here — use leo:worktrees' raw `git worktree` fallback for "
+        "\nSkills are invoked by bare name here — `brainstorming`, `verification` — not with the "
+        "`leo:` prefix the policy above uses. The plugin registers the skills directory by path, so "
+        "OpenCode names each skill from its own `name:` frontmatter and applies no plugin "
+        "namespace. Read `leo:<skill>` anywhere in the policy as `<skill>` on this harness.\n"
+        "\nNo `EnterWorktree` tool exists here — use the worktrees skill's raw `git worktree` fallback for "
         "isolated branch work. State reads and writes go through `python3 <plugin-root>/scripts/state.py` "
         "(`get` / `merge` / `path`), same contract as every other harness. There is no Workflow tool "
         "and no `cost-tiered-fix.js` here — a batch of independent tasks is fanned out as manual "
@@ -262,6 +266,50 @@ def _mapping_docs(config):
     }
 
 
+def _payload_readme(config):
+    """The npm landing page for `leos-agent`.
+
+    The root README cannot be symlinked in: every harness copies or caches
+    this payload on its own, so a link pointing outside it dangles, and the
+    packaging tests forbid symlinks here for exactly that reason. Generating
+    it instead keeps one source and lets --check catch drift.
+
+    Scoped to OpenCode deliberately — npm is only how OpenCode installs this.
+    Every other harness has a native marketplace, and its instructions would
+    be noise on this page.
+    """
+    opencode = config["harnesses"]["opencode"]
+    return (
+        GENERATED
+        + "\n# Leo's Agent\n\n"
+        "Leo's Agent is a portable agent operating policy: cost-tiered model routing, specialist "
+        "subagent roles, process skills, execute-then-review discipline, and a narrow "
+        "catastrophic-command guard.\n\n"
+        "This npm package is the **OpenCode** distribution. Claude Code, Codex, Cursor, and Hermes "
+        "each install it through their own plugin system — see "
+        "[the repository](https://github.com/foxhatleo/leos-agent) for those.\n\n"
+        "## Install\n\n"
+        "Add it to `~/.config/opencode/opencode.json`:\n\n"
+        '```json\n{ "$schema": "https://opencode.ai/config.json", "plugin": ["leos-agent"] }\n```\n\n'
+        "Start a new OpenCode session. The plugin registers the skills directory, the "
+        f"{len(json.loads(_opencode_agents(config)))} subagent roles, and the operating policy, and installs the "
+        "bash deletion tripwire.\n\n"
+        "If the skills do not appear, register the path by hand — npm installs into OpenCode's own "
+        "cache and the location varies by version:\n\n"
+        '```json\n{ "skills": { "paths": ["~/.cache/opencode/node_modules/leos-agent/skills"] } }\n```\n\n'
+        "## Model tiers\n\n"
+        "Tier names describe the kind of work, not a fixed provider model.\n\n"
+        + _table({tier: opencode[tier] for tier in ("fable", "opus", "sonnet", "haiku")})
+        + "\n\nFable and Opus collapse onto one model here, so `expert` is not registered as an "
+        "agent and escalation caps at Opus. Retier by editing `config/models.json` and re-running "
+        "`scripts/render_adapters.py`.\n\n"
+        "## Links\n\n"
+        "- [Repository and full documentation](https://github.com/foxhatleo/leos-agent)\n"
+        "- [Operating policy](https://github.com/foxhatleo/leos-agent/blob/main/plugins/leo/skills/using-leo/SKILL.md)\n\n"
+        "MIT licensed.\n"
+    )
+
+
 def render(config):
     outputs = {}
     for role, tier in sorted(config["roles"].items()):
@@ -274,6 +322,7 @@ def render(config):
     for harness, content in _mapping_docs(config).items():
         outputs[ROOT / "skills" / "using-leo" / "references" / f"{harness}-mapping.md"] = content
     outputs[ROOT / "adapters" / "opencode" / "agents.json"] = _opencode_agents(config)
+    outputs[ROOT / "README.md"] = _payload_readme(config)
 
     # The manifest is no longer rewritten here: per-install model overrides
     # were retired along with the placeholder they fed. Retiering means editing
@@ -305,6 +354,7 @@ def main():
         "agents/*.md",
         "adapters/cursor/agents/*.md",
         "adapters/opencode/agents.json",
+        "README.md",
         "skills/using-leo/references/*-mapping.md",
     ):
         for path in sorted(ROOT.glob(pattern)):

package/skills/using-leo/references/opencode-mapping.md

@@ -14,7 +14,9 @@ Roles register as native OpenCode agents (from `adapters/opencode/agents.json`,
 
 Read-only is harness-enforced here, unlike Codex and Cursor: read-only roles carry a generated `permission.edit: deny`, so an off-policy write attempt is refused by OpenCode itself, not merely discouraged by the prompt. Write-capable agents additionally carry coarse `rm -rf` bash denies as a stopgap for opencode#5894 (unconfirmed whether `tool.execute.before` also intercepts subagent bash); the precise tripwire stays `hooks/bash-guard.py` on the primary agent.
 
-No `EnterWorktree` tool exists here — use leo:worktrees' raw `git worktree` fallback for isolated branch work. State reads and writes go through `python3 <plugin-root>/scripts/state.py` (`get` / `merge` / `path`), same contract as every other harness. There is no Workflow tool and no `cost-tiered-fix.js` here — a batch of independent tasks is fanned out as manual parallel task-tool subagent spawns instead.
+Skills are invoked by bare name here — `brainstorming`, `verification` — not with the `leo:` prefix the policy above uses. The plugin registers the skills directory by path, so OpenCode names each skill from its own `name:` frontmatter and applies no plugin namespace. Read `leo:<skill>` anywhere in the policy as `<skill>` on this harness.
+
+No `EnterWorktree` tool exists here — use the worktrees skill's raw `git worktree` fallback for isolated branch work. State reads and writes go through `python3 <plugin-root>/scripts/state.py` (`get` / `merge` / `path`), same contract as every other harness. There is no Workflow tool and no `cost-tiered-fix.js` here — a batch of independent tasks is fanned out as manual parallel task-tool subagent spawns instead.
 
 Tier collapse here: Fable≡Opus (`moonshotai/kimi-k3`), Sonnet≡Haiku (`z-ai/glm-5.2`) — routing between collapsed rungs buys role, not power. Fable is not a real rung: `expert` cannot break a deadlock a collapsed Opus already lost, so cap escalation at Opus and report.