@dev-sergeev/datalab-kit 0.1.5 → 0.1.7

package/skills/skill-creator/references/schemas.md

@@ -0,0 +1,521 @@
+# JSON Schemas
+
+> Modified for Datalab Kit/Pi from Anthropic's skill-creator,
+> revision `dbd4588f9e1033efb41dad4bef2f7947c8993d44`.
+> Changes: Pi telemetry, writable workspace layout, and evaluation field guidance.
+> See [Apache-2.0 license](../LICENSE.txt) and [source attribution](../SKILL.md).
+
+## Contents
+
+- [Workspace layout](#workspace-layout)
+- [evals.json](#evalsjson)
+- [eval_metadata.json](#eval_metadatajson)
+- [history.json](#historyjson)
+- [grading.json](#gradingjson)
+- [metrics.json](#metricsjson)
+- [timing.json](#timingjson)
+- [benchmark.json](#benchmarkjson)
+- [comparison.json](#comparisonjson)
+- [analysis.json](#analysisjson)
+
+## Workspace layout
+
+Use a writable workspace outside the installed skill-creator package:
+
+```text
+<workspace>/
+  candidate/<skill-name>/SKILL.md
+  skill-snapshot/<skill-name>/SKILL.md
+  iteration-1/
+    eval-1-descriptive-name/
+      eval_metadata.json
+      with_skill/run-1/
+        transcript.jsonl
+        stderr.log
+        timing.json
+        grading.json
+        outputs/
+      without_skill/run-1/
+        transcript.jsonl
+        stderr.log
+        timing.json
+        grading.json
+        outputs/
+    benchmark.json
+    benchmark.md
+    feedback.json
+```
+
+Repeat under `run-2/`, `run-3/`, etc. for variance measurements; `run-1/` is
+required even for a single run because aggregation searches `run-*` directories.
+For an update, `without_skill` contains the old-skill snapshot execution; record
+that baseline identity explicitly in metadata and benchmark notes. It does not
+mean that the old-skill arm lacked a skill.
+
+Examples below illustrate values, not telemetry that every Pi provider supplies.
+Omit unavailable measurements; never substitute character counts for tokens or
+zero for missing usage. Keep executor usage/time separate from grading and
+orchestration. Use absolute paths in executor prompts; keep raw Pi JSONL for
+auditing and derive metrics only from complete events.
+
+This document defines the JSON schemas used by skill-creator.
+
+---
+
+## evals.json
+
+Defines the evals for a skill. Located at `evals/evals.json` within the writable candidate skill, not the installed creator package.
+
+```json
+{
+  "skill_name": "example-skill",
+  "evals": [
+    {
+      "id": 1,
+      "prompt": "User's example prompt",
+      "expected_output": "Description of expected result",
+      "files": ["evals/files/sample1.pdf"],
+      "expectations": [
+        "The output includes X",
+        "The skill used script Y"
+      ]
+    }
+  ]
+}
+```
+
+**Fields:**
+- `skill_name`: Name matching the skill's frontmatter
+- `evals[].id`: Unique integer identifier
+- `evals[].prompt`: The task to execute
+- `evals[].expected_output`: Human-readable description of success
+- `evals[].files`: Optional list of input file paths (relative to skill root)
+- `evals[].expectations`: List of verifiable statements
+
+The main workflow calls these statements "assertions". Store them as
+`expectations` here and copy the strings into `assertions` in eval_metadata.json.
+
+---
+
+## eval_metadata.json
+
+Located at `<workspace>/iteration-N/eval-ID-name/eval_metadata.json`.
+
+```json
+{
+  "eval_id": 1,
+  "eval_name": "table-extraction",
+  "prompt": "Extract the line items from the supplied invoice into a CSV.",
+  "assertions": ["Every invoice line item is present in the CSV"],
+  "baseline": "original skill snapshot"
+}
+```
+
+`eval_id`, `eval_name`, `prompt`, and `assertions` support the viewer and grading
+workflow. `baseline` is optional human-readable context: use `"no skill"` for a
+new skill or identify the snapshot used for an update. Keep this metadata outside
+executor outputs so the executor cannot accidentally consume expected answers.
+
+---
+
+## history.json
+
+Tracks version progression in Improve mode. Located at workspace root.
+
+```json
+{
+  "started_at": "2026-01-15T10:30:00Z",
+  "skill_name": "pdf",
+  "current_best": "v2",
+  "iterations": [
+    {
+      "version": "v0",
+      "parent": null,
+      "expectation_pass_rate": 0.65,
+      "grading_result": "baseline",
+      "is_current_best": false
+    },
+    {
+      "version": "v1",
+      "parent": "v0",
+      "expectation_pass_rate": 0.75,
+      "grading_result": "won",
+      "is_current_best": false
+    },
+    {
+      "version": "v2",
+      "parent": "v1",
+      "expectation_pass_rate": 0.85,
+      "grading_result": "won",
+      "is_current_best": true
+    }
+  ]
+}
+```
+
+**Fields:**
+- `started_at`: ISO timestamp of when improvement started
+- `skill_name`: Name of the skill being improved
+- `current_best`: Version identifier of the best performer
+- `iterations[].version`: Version identifier (v0, v1, ...)
+- `iterations[].parent`: Parent version this was derived from
+- `iterations[].expectation_pass_rate`: Pass rate from grading
+- `iterations[].grading_result`: "baseline", "won", "lost", or "tie"
+- `iterations[].is_current_best`: Whether this is the current best version
+
+---
+
+## grading.json
+
+Output from the grader agent. Located at `<run-dir>/grading.json`.
+
+```json
+{
+  "expectations": [
+    {
+      "text": "The output includes the name 'John Smith'",
+      "passed": true,
+      "evidence": "Found in transcript Step 3: 'Extracted names: John Smith, Sarah Johnson'"
+    },
+    {
+      "text": "The spreadsheet has a SUM formula in cell B10",
+      "passed": false,
+      "evidence": "No spreadsheet was created. The output was a text file."
+    }
+  ],
+  "summary": {
+    "passed": 2,
+    "failed": 1,
+    "total": 3,
+    "pass_rate": 0.67
+  },
+  "execution_metrics": {
+    "tool_calls": {
+      "read": 5,
+      "write": 2,
+      "bash": 8
+    },
+    "total_tool_calls": 15,
+    "total_steps": 6,
+    "errors_encountered": 0,
+    "output_chars": 12450,
+    "transcript_chars": 3200
+  },
+  "timing": {
+    "executor_duration_seconds": 165.0,
+    "grader_duration_seconds": 26.0,
+    "total_duration_seconds": 165.0
+  },
+  "claims": [
+    {
+      "claim": "The form has 12 fillable fields",
+      "type": "factual",
+      "verified": true,
+      "evidence": "Counted 12 fields in field_info.json"
+    }
+  ],
+  "user_notes_summary": {
+    "uncertainties": ["Used 2023 data, may be stale"],
+    "needs_review": [],
+    "workarounds": ["Fell back to text overlay for non-fillable fields"]
+  },
+  "eval_feedback": {
+    "suggestions": [
+      {
+        "assertion": "The output includes the name 'John Smith'",
+        "reason": "A hallucinated document that mentions the name would also pass"
+      }
+    ],
+    "overall": "Assertions check presence but not correctness."
+  }
+}
+```
+
+**Fields:**
+- `expectations[]`: Graded expectations with evidence
+- `summary`: Aggregate pass/fail counts
+- `execution_metrics`: Optional recorded tool usage and output size (from executor's metrics.json or measured JSONL events); omit when unavailable
+- `timing`: Optional measured executor wall time (from timing.json); do not include grader/orchestrator usage in benchmark time
+- `claims`: Extracted and verified claims from the output
+- `user_notes_summary`: Issues flagged by the executor
+- `eval_feedback`: (optional) Improvement suggestions for the evals, only present when the grader identifies issues worth raising
+
+---
+
+## metrics.json
+
+Output from the executor agent. Located at `<run-dir>/outputs/metrics.json`.
+
+```json
+{
+  "tool_calls": {
+    "read": 5,
+    "write": 2,
+    "bash": 8,
+    "edit": 1,
+    "find": 2,
+    "grep": 0
+  },
+  "total_tool_calls": 18,
+  "total_steps": 6,
+  "files_created": ["filled_form.pdf", "field_values.json"],
+  "errors_encountered": 0,
+  "output_chars": 12450,
+  "transcript_chars": 3200
+}
+```
+
+**Fields:**
+- `tool_calls`: Count per tool type
+- `total_tool_calls`: Sum of all tool calls
+- `total_steps`: Number of major execution steps
+- `files_created`: List of output files created
+- `errors_encountered`: Number of errors during execution
+- `output_chars`: Total character count of output files
+- `transcript_chars`: Character count of transcript
+
+Pi emits `tool_execution_start`/`tool_execution_end` events with `toolName`,
+`toolCallId`, and (on completion) `isError`. Count each executed call once.
+`total_steps` is an optional workflow-level count, not a native Pi metric.
+File sizes and created-file lists can be measured from the outputs; they are not
+automatically emitted by Pi. Include only measurements actually collected.
+
+---
+
+## timing.json
+
+Wall clock timing for a run. Located at `<run-dir>/timing.json`.
+
+**How to capture:** Measure the fresh executor Pi process using a monotonic clock. For tokens, sum `message.usage.totalTokens` from completed assistant `message_end` events in that executor's JSONL, once per message. Preserve the JSONL so this can be audited or recovered. Do not use rounded subagent summaries or assume completion notifications contain numeric `total_tokens`/`duration_ms`. Pi subagent lifetime usage is a different accounting surface and can include orchestration; do not mix it into executor measurements.
+
+```json
+{
+  "total_tokens": 84852,
+  "duration_ms": 23332,
+  "total_duration_seconds": 23.332,
+  "executor_start": "2026-01-15T10:30:00Z",
+  "executor_end": "2026-01-15T10:30:23.332Z",
+  "executor_duration_seconds": 23.332
+}
+```
+
+`total_tokens` is optional when the provider supplies no usable token telemetry.
+The start/end timestamps are optional context; elapsed durations should come
+from a monotonic clock. If separately measuring grading, use explicitly named
+grader fields, but keep `total_duration_seconds` as executor time for benchmark
+compatibility. A provider-reported zero is not the same as missing telemetry.
+
+---
+
+## benchmark.json
+
+Output from benchmark aggregation. Located at `<workspace>/iteration-N/benchmark.json` (or the explicitly selected benchmark directory).
+
+```json
+{
+  "metadata": {
+    "skill_name": "pdf",
+    "skill_path": "/path/to/pdf",
+    "executor_model": "provider/model-id",
+    "analyzer_model": "most-capable-model",
+    "timestamp": "2026-01-15T10:30:00Z",
+    "evals_run": [1, 2, 3],
+    "runs_per_configuration": 3
+  },
+
+  "runs": [
+    {
+      "eval_id": 1,
+      "eval_name": "Ocean",
+      "configuration": "with_skill",
+      "run_number": 1,
+      "result": {
+        "pass_rate": 0.85,
+        "passed": 6,
+        "failed": 1,
+        "total": 7,
+        "time_seconds": 42.5,
+        "tokens": 3800,
+        "tool_calls": 18,
+        "errors": 0
+      },
+      "expectations": [
+        {"text": "...", "passed": true, "evidence": "..."}
+      ],
+      "notes": [
+        "Used 2023 data, may be stale",
+        "Fell back to text overlay for non-fillable fields"
+      ]
+    }
+  ],
+
+  "run_summary": {
+    "with_skill": {
+      "pass_rate": {"mean": 0.85, "stddev": 0.05, "min": 0.80, "max": 0.90},
+      "time_seconds": {"mean": 45.0, "stddev": 12.0, "min": 32.0, "max": 58.0},
+      "tokens": {"mean": 3800, "stddev": 400, "min": 3200, "max": 4100}
+    },
+    "without_skill": {
+      "pass_rate": {"mean": 0.35, "stddev": 0.08, "min": 0.28, "max": 0.45},
+      "time_seconds": {"mean": 32.0, "stddev": 8.0, "min": 24.0, "max": 42.0},
+      "tokens": {"mean": 2100, "stddev": 300, "min": 1800, "max": 2500}
+    },
+    "delta": {
+      "pass_rate": "+0.50",
+      "time_seconds": "+13.0",
+      "tokens": "+1700"
+    }
+  },
+
+  "notes": [
+    "Assertion 'Output is a PDF file' passes 100% in both configurations - may not differentiate skill value",
+    "Eval 3 shows high variance (50% ± 40%) - may be flaky or model-dependent",
+    "Without-skill runs consistently fail on table extraction expectations",
+    "Skill adds 13s average execution time but improves pass rate by 50%"
+  ]
+}
+```
+
+**Fields:**
+- `metadata`: Information about the benchmark run
+  - `skill_name`: Name of the skill
+  - `timestamp`: When the benchmark was run
+  - `evals_run`: List of eval names or IDs
+  - `runs_per_configuration`: Observed repetitions per evaluation/configuration pair; null when counts differ or are unavailable. Per-run records remain authoritative.
+- `runs[]`: Individual run results
+  - `eval_id`: Numeric eval identifier
+  - `eval_name`: Human-readable eval name (used as section header in the viewer)
+  - `configuration`: Use `"with_skill"` or `"without_skill"` for the documented comparison workflow (the viewer uses these names for grouping and color coding). An old-skill baseline still uses `"without_skill"` with its real identity explained in notes.
+  - `run_number`: Integer run number (1, 2, 3...)
+  - `result`: Nested object with graded `pass_rate`, `passed`, `total`; `time_seconds`, `tokens`, tool counts, and errors are measured numbers or `null` when unavailable
+- `run_summary`: Statistical aggregates per configuration
+  - `with_skill` / `without_skill`: Each contains metric objects with `mean`, `stddev`, `min`, and `max`; absent measurements produce null statistic values, not zero usage
+  - `delta`: Difference strings like `"+0.50"`, `"+13.0"`, `"+1700"`, or `null` when unavailable
+- `notes`: Freeform observations from the analyzer
+
+**Important:** The viewer reads these field names exactly. Using `config` instead of `configuration`, or putting `pass_rate` at the top level of a run instead of nested under `result`, will cause the viewer to show empty/zero values. Always reference this schema when generating benchmark.json manually.
+
+---
+
+## comparison.json
+
+Output from blind comparator. Located at `<grading-dir>/comparison-N.json`.
+
+```json
+{
+  "winner": "A",
+  "reasoning": "Output A provides a complete solution with proper formatting and all required fields. Output B is missing the date field and has formatting inconsistencies.",
+  "rubric": {
+    "A": {
+      "content": {
+        "correctness": 5,
+        "completeness": 5,
+        "accuracy": 4
+      },
+      "structure": {
+        "organization": 4,
+        "formatting": 5,
+        "usability": 4
+      },
+      "content_score": 4.7,
+      "structure_score": 4.3,
+      "overall_score": 9.0
+    },
+    "B": {
+      "content": {
+        "correctness": 3,
+        "completeness": 2,
+        "accuracy": 3
+      },
+      "structure": {
+        "organization": 3,
+        "formatting": 2,
+        "usability": 3
+      },
+      "content_score": 2.7,
+      "structure_score": 2.7,
+      "overall_score": 5.4
+    }
+  },
+  "output_quality": {
+    "A": {
+      "score": 9,
+      "strengths": ["Complete solution", "Well-formatted", "All fields present"],
+      "weaknesses": ["Minor style inconsistency in header"]
+    },
+    "B": {
+      "score": 5,
+      "strengths": ["Readable output", "Correct basic structure"],
+      "weaknesses": ["Missing date field", "Formatting inconsistencies", "Partial data extraction"]
+    }
+  },
+  "expectation_results": {
+    "A": {
+      "passed": 4,
+      "total": 5,
+      "pass_rate": 0.80,
+      "details": [
+        {"text": "Output includes name", "passed": true}
+      ]
+    },
+    "B": {
+      "passed": 3,
+      "total": 5,
+      "pass_rate": 0.60,
+      "details": [
+        {"text": "Output includes name", "passed": true}
+      ]
+    }
+  }
+}
+```
+
+---
+
+## analysis.json
+
+Output from post-hoc analyzer. Located at `<grading-dir>/analysis.json`.
+
+```json
+{
+  "comparison_summary": {
+    "winner": "A",
+    "winner_skill": "path/to/winner/skill",
+    "loser_skill": "path/to/loser/skill",
+    "comparator_reasoning": "Brief summary of why comparator chose winner"
+  },
+  "winner_strengths": [
+    "Clear step-by-step instructions for handling multi-page documents",
+    "Included validation script that caught formatting errors"
+  ],
+  "loser_weaknesses": [
+    "Vague instruction 'process the document appropriately' led to inconsistent behavior",
+    "No script for validation, agent had to improvise"
+  ],
+  "instruction_following": {
+    "winner": {
+      "score": 9,
+      "issues": ["Minor: skipped optional logging step"]
+    },
+    "loser": {
+      "score": 6,
+      "issues": [
+        "Did not use the skill's formatting template",
+        "Invented own approach instead of following step 3"
+      ]
+    }
+  },
+  "improvement_suggestions": [
+    {
+      "priority": "high",
+      "category": "instructions",
+      "suggestion": "Replace 'process the document appropriately' with explicit steps",
+      "expected_impact": "Would eliminate ambiguity that caused inconsistent behavior"
+    }
+  ],
+  "transcript_insights": {
+    "winner_execution_pattern": "Read skill -> Followed 5-step process -> Used validation script",
+    "loser_execution_pattern": "Read skill -> Unclear on approach -> Tried 3 different methods"
+  }
+}
+```