opencode-harness-kit 0.4.3 → 0.4.4

package/profiles/fleet-all.json

@@ -43,6 +43,10 @@
     "sweep": "none",
     "longform": null
   },
+  "utility": {
+    "model": "deepseek-v4.1-flash-free",
+    "why": "Cheapest route measured for a short auxiliary call: 49 reasoning tokens and 7.2s for a title, against 110 and 13.4s on big-pickle and 332 and 10.6s on nemotron-3.5-super-vl. It is in seven of this fleet's ten chains, so a failover has a tier-capable model to fall back to."
+  },
   "roleModels": {
     "build": {
       "model": "minimax-m3.1-flash-free",
@@ -73,7 +77,8 @@
       "chain": [
         "nemotron-3.5-lightning-free",
         "big-pickle",
-        "longcat-2.5-preview-free"
+        "longcat-2.5-preview-free",
+        "deepseek-v4.1-flash-free"
       ]
     },
     "verify": {

package/profiles/fleet-safe.json

@@ -40,9 +40,13 @@
     "vision": null,
     "doc": null,
     "schema": "max",
-    "sweep": "none",
+    "sweep": "low",
     "longform": null
   },
+  "utility": {
+    "model": "deepseek-v4.1-flash-free",
+    "why": "Cheapest route measured for a short auxiliary call: 49 reasoning tokens and 7.2s for a title, against 110 and 13.4s on big-pickle and 332 and 10.6s on nemotron-3.5-super-vl. It is also in every chain of this fleet, so a failover cannot leave the utility route pointing at a model that is not in the fleet."
+  },
   "roleModels": {
     "build": {
       "model": "minimax-m3.1-flash-free",
@@ -109,10 +113,10 @@
       ]
     },
     "sweep": {
-      "model": "longcat-2.5-preview-free",
+      "model": "deepseek-v4.1-flash-free",
       "chain": [
-        "longcat-2.5-preview-free",
         "deepseek-v4.1-flash-free",
+        "longcat-2.5-preview-free",
         "minimax-m3.1-flash-free"
       ]
     },

package/profiles/schema.json

@@ -1,17 +1,31 @@
 {
   "$schema": "http://json-schema.org/draft-07/schema#",
-  "$id": "https://github.com/anomalyco/opencode-harness-kit/profiles/schema.json",
+  "$id": "https://github.com/pham-canh/space-harness/profiles/schema.json",
   "title": "OpenCode harness profile",
   "description": "Everything that changes when you change model. Plugins read this file and contain no model names of their own.",
   "type": "object",
-  "required": ["id", "provider", "model", "retention", "limits", "effort", "roles"],
+  "required": [
+    "id",
+    "provider",
+    "model",
+    "retention",
+    "limits",
+    "effort",
+    "roles"
+  ],
   "additionalProperties": false,
   "properties": {
+    "$comment": {
+      "type": "string",
+      "description": "Free text for a human. Allowed here because every shipped profile carries one, and lanes.schema.json declares it too."
+    },
     "id": {
       "type": "string",
       "description": "Slug used as the profile filename and the oc-harness use argument."
     },
-    "title": { "type": "string" },
+    "title": {
+      "type": "string"
+    },
     "provider": {
       "type": "string",
       "description": "opencode provider id, e.g. opencode, opencode-go, anthropic."
@@ -21,7 +35,10 @@
       "description": "Model id without the provider prefix, e.g. beam-free."
     },
     "family": {
-      "type": ["string", "null"],
+      "type": [
+        "string",
+        "null"
+      ],
       "description": "Model family for grouping. Stealth models report null upstream, so this may be set by hand."
     },
     "verified": {
@@ -29,23 +46,46 @@
       "description": "ISO date a human checked this profile against live metadata."
     },
     "retention": {
-      "enum": ["zero-retention", "retained", "may-train", "unknown"],
+      "enum": [
+        "zero-retention",
+        "retained",
+        "may-train",
+        "unknown"
+      ],
       "description": "Must match an entry in retention.yml. A profile whose model is unknown there cannot become the default."
     },
     "maxRetention": {
-      "enum": ["zero-retention", "retained", "may-train"],
+      "enum": [
+        "zero-retention",
+        "retained",
+        "may-train"
+      ],
       "description": "The worst retention a lane in this profile is allowed to resolve to. A lane walks its model chain and takes the first rung at or below this ceiling, so one profile can serve a fleet without sending a prompt to a provider that trains on it. Defaults to the profile's own retention."
     },
-    "notes": { "type": "string" },
+    "notes": {
+      "type": "string"
+    },
     "limits": {
       "type": "object",
-      "required": ["context", "output"],
+      "required": [
+        "context",
+        "output"
+      ],
       "additionalProperties": false,
       "description": "Token ceilings from the provider catalog. `input` is absent for several free models, in which case the loader derives it from `context`.",
       "properties": {
-        "context": { "type": "integer", "minimum": 1 },
-        "input": { "type": "integer", "minimum": 1 },
-        "output": { "type": "integer", "minimum": 1 }
+        "context": {
+          "type": "integer",
+          "minimum": 1
+        },
+        "input": {
+          "type": "integer",
+          "minimum": 1
+        },
+        "output": {
+          "type": "integer",
+          "minimum": 1
+        }
       }
     },
     "supports": {
@@ -53,10 +93,18 @@
       "additionalProperties": false,
       "description": "Capability flags. Missing or false means do not emit the related config.",
       "properties": {
-        "reasoning": { "type": "boolean" },
-        "temperature": { "type": "boolean" },
-        "toolCall": { "type": "boolean" },
-        "structuredOutput": { "type": "boolean" },
+        "reasoning": {
+          "type": "boolean"
+        },
+        "temperature": {
+          "type": "boolean"
+        },
+        "toolCall": {
+          "type": "boolean"
+        },
+        "structuredOutput": {
+          "type": "boolean"
+        },
         "attachment": {
           "type": "boolean",
           "description": "Accepts image or video input. Gates screenshot-based verification."
@@ -69,10 +117,26 @@
       "properties": {
         "levels": {
           "type": "array",
-          "items": { "type": "string" },
-          "examples": [["low", "medium", "high", "xhigh", "max"]]
+          "items": {
+            "type": "string"
+          },
+          "examples": [
+            [
+              "low",
+              "medium",
+              "high",
+              "xhigh",
+              "max"
+            ]
+          ]
+        },
+        "default": {
+          "type": [
+            "string",
+            "null"
+          ],
+          "description": "The level the session runs at by default. Null means the model has no tunable effort and no variant is emitted."
         },
-        "default": { "type": "string" },
         "reserveShare": {
           "type": "number",
           "minimum": 0,
@@ -84,19 +148,57 @@
     },
     "roles": {
       "type": "object",
-      "description": "Requested effort per lane. Resolved against effort.levels, never emitted raw. A lane listed here is enabled; a lane absent from here is not installed.",
+      "description": "Requested effort per lane, never emitted raw. A lane listed here is enabled; a lane absent from here is not installed. It is resolved against the levels of the model the lane actually runs on, which is the profile's own effort.levels for a single-model profile and the lane model's levels for a fleet. `none` means reasoning off: a model that does not publish it gets no variant at all.",
       "additionalProperties": false,
       "properties": {
-        "build": { "$ref": "#/definitions/effortRequest" },
-        "plan": { "$ref": "#/definitions/effortRequest" },
-        "reviewer": { "$ref": "#/definitions/effortRequest" },
-        "locate": { "$ref": "#/definitions/effortRequest" },
-        "verify": { "$ref": "#/definitions/effortRequest" },
-        "vision": { "$ref": "#/definitions/effortRequest" },
-        "doc": { "$ref": "#/definitions/effortRequest" },
-        "schema": { "$ref": "#/definitions/effortRequest" },
-        "sweep": { "$ref": "#/definitions/effortRequest" },
-        "longform": { "$ref": "#/definitions/effortRequest" }
+        "build": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "plan": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "reviewer": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "locate": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "verify": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "vision": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "doc": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "schema": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "sweep": {
+          "$ref": "#/definitions/effortRequest"
+        },
+        "longform": {
+          "$ref": "#/definitions/effortRequest"
+        }
+      }
+    },
+    "utility": {
+      "type": "object",
+      "description": "The route opencode uses for its short auxiliary calls - a title, a one-line summary. Optional: a profile without one keeps the session model, which is what a single-model profile wants, because 'one route, and it is the one you asked for' is a promise those profiles make.",
+      "additionalProperties": false,
+      "required": [
+        "model"
+      ],
+      "properties": {
+        "model": {
+          "type": "string",
+          "description": "Model id without the provider prefix. Must be a model this profile already resolves to, so the utility route is never a model the fleet cannot run. The route itself does not change on failover; the profile keeps it."
+        },
+        "why": {
+          "type": "string",
+          "description": "The measurement that chose it. A route chosen by taste is a route that drifts."
+        }
       }
     },
     "roleModels": {
@@ -104,16 +206,36 @@
       "description": "Model per lane. A lane absent from `roles` is not enabled. A lane absent from `roleModels` runs on the profile's base model.",
       "additionalProperties": false,
       "properties": {
-        "build": { "$ref": "#/definitions/roleModel" },
-        "plan": { "$ref": "#/definitions/roleModel" },
-        "reviewer": { "$ref": "#/definitions/roleModel" },
-        "locate": { "$ref": "#/definitions/roleModel" },
-        "verify": { "$ref": "#/definitions/roleModel" },
-        "vision": { "$ref": "#/definitions/roleModel" },
-        "doc": { "$ref": "#/definitions/roleModel" },
-        "schema": { "$ref": "#/definitions/roleModel" },
-        "sweep": { "$ref": "#/definitions/roleModel" },
-        "longform": { "$ref": "#/definitions/roleModel" }
+        "build": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "plan": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "reviewer": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "locate": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "verify": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "vision": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "doc": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "schema": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "sweep": {
+          "$ref": "#/definitions/roleModel"
+        },
+        "longform": {
+          "$ref": "#/definitions/roleModel"
+        }
       }
     },
     "prune": {
@@ -121,7 +243,9 @@
       "additionalProperties": false,
       "description": "Deterministic context pruning. Only used when a model-driven compressor is unavailable or unproven.",
       "properties": {
-        "enabled": { "type": "boolean" },
+        "enabled": {
+          "type": "boolean"
+        },
         "keepRecentToolCalls": {
           "type": "integer",
           "minimum": 1,
@@ -137,14 +261,16 @@
         "keepUserMessages": {
           "type": "boolean",
           "default": true,
-          "description": "Never stub a user message. Their instructions are the point of the session."
+          "description": "Never stub anything inside a user message. Their instructions are the point of the session. This covers a tool result a user attached as well as their text: user text is kept either way, because prose is kept everywhere."
         }
       }
     },
     "fallback": {
       "type": "array",
       "description": "Ordered profile ids to fall back to. Selection happens at session start, not mid-session: no opencode plugin hook can change a live session's model.",
-      "items": { "type": "string" }
+      "items": {
+        "type": "string"
+      }
     }
   },
   "definitions": {
@@ -152,28 +278,33 @@
       "type": "object",
       "description": "A model for one lane, plus the effort levels that model actually offers, plus the ordered list of models that may serve the lane when the first one is spent or too private.",
       "additionalProperties": false,
-      "required": ["model"],
+      "required": [
+        "model"
+      ],
       "properties": {
         "model": {
           "type": "string",
           "description": "Model id without the provider prefix. Must resolve to a profile in this directory, so its limits and retention are known."
         },
-        "levels": {
-          "type": "array",
-          "items": { "type": "string" },
-          "description": "Effort levels this role's model offers. Omitted means the role emits no variant at all, which is the right reading for a model with no tunable effort."
-        },
         "chain": {
           "type": "array",
-          "items": { "type": "string" },
+          "items": {
+            "type": "string"
+          },
           "description": "Ordered models that may serve this lane, best first. The first rung the profile's maxRetention ceiling allows is the one that runs it, and the router walks the same chain when a route is spent."
         }
       }
     },
     "effortRequest": {
       "oneOf": [
-        { "type": "string", "description": "Desired level, resolved to the nearest available one." },
-        { "type": "null", "description": "This role should not set a variant at all." }
+        {
+          "type": "string",
+          "description": "Desired level, resolved to the nearest available one."
+        },
+        {
+          "type": "null",
+          "description": "This role should not set a variant at all."
+        }
       ]
     }
   }

package/prompts/build.md

@@ -17,8 +17,10 @@ You implement one task from a plan. One. Not the whole plan.
   cannot complete the task, say so before writing a partial one.
 - Never widen the change. A drive-by refactor in the same file is a separate
   task; do not report it as part of this one.
-- Ask before `git push`, publish, or opening a pull request. Committing is
-  allowed but only when the task asked for it.
+- Committing and pushing the branch you were asked to finish is allowed, and
+  you say what you pushed. Ask before publishing a package, opening a pull
+  request, or changing repository visibility. Committing is allowed but only
+  when the task asked for it.
 - Never fabricate a passing result. A failing test reported honestly is worth
   more than a green one you did not get.
 - Never re-run a command just to get its output back into context. Say what you

package/skills/space-build/SKILL.md

@@ -25,8 +25,10 @@ description: Implement one planned task and verify it with a real command. Use w
    Fix the cause, then run it again. The third identical attempt is a signal to
    stop and report.
 7. **Mark the task done in `.space/tasks.md`** and paste the verify command's
-   verbatim final lines into its `verify:` line. A task marked done without
-   that has not been verified.
+   verbatim final lines into its `output:` line. The `verify:` line keeps the
+   command and nothing else - it is what the driver runs, so a `verify:` that
+   reads `npm test - 3 passed` is a command that does not exist. A task marked
+   done without the output has not been verified.
 8. **Report the command and its verbatim output.**
 
 ## Test-first when the plan called for it

package/skills/space-finish/SKILL.md

@@ -37,8 +37,12 @@ over an unreviewed or blocked change is the failure this skill exists to stop.
 
 ## The publish boundary
 
-Publishing needs human approval. Ask before `git push`, opening a pull request,
-publishing a package, or changing repository visibility, even if the task
+Pushing the branch you were asked to finish is not one of the decisions that
+need a human: commit it, push it, and say what you pushed.
+
+Everything outward-facing past the push is. Ask before publishing a package,
+opening a pull request, changing repository visibility, force-pushing,
+rewriting published history, or deleting a branch or tag - even if the task
 seems to imply it. Permissions default to ask so opencode prompts first, but
 the boundary is yours to hold, not only the hook's.
 

package/skills/space-plan/reference/task-template.md

@@ -20,15 +20,23 @@ has been pruned. Never keep it in your head.
 - [ ] <task, small enough to finish in a few minutes>
   - files: <exact paths to change>
   - change: <what the change is, concretely, not "handle errors">
-  - verify: <the exact command, and what passing looks like>
-  - status: pending | done | blocked
+  - verify: <the exact command that proves it, and nothing else>
+  - status: <pending | done | blocked>
+  - output: <the command's verbatim final lines>
 - [x] <finished task>
   - files: <paths>
   - change: <what was done, including anything the original task did not name>
-  - verify: <the command that was actually run, and its verbatim final lines>
+  - verify: <the command that was actually run>
+  - output: <the command's verbatim final lines>
   - status: done
 ```
 
+`verify:` is a command and nothing else, because the driver runs it. The
+output of that command goes on the `output:` line under it, never on the
+`verify:` line and never appended to it: a `verify:` that reads
+`npm test - 3 passed` is a command that does not exist, and the loop would
+fail on the command rather than on the work.
+
 How the loop uses it:
 
 | step | does |
@@ -41,7 +49,7 @@ How the loop uses it:
 Rules:
 
 - One file per body of work. Start a new one only when the goal changes.
-- A task is `done` only with the verify command's real output pasted into its
-  `verify:` line. A task marked done without it has not been verified.
+- A task is `done` only with the verify command's real output in its `output:`
+  line. A task marked done without it has not been verified.
 - Add a Decisions line the moment something was chosen rather than asked. That
   is the line that stops the next session re-litigating it.