back
loading skill details...
Score a generated CLI against the Steinberger bar, compare two CLIs side-by-side
/printing-press-score
Score generated CLIs against the Steinberger bar. Supports rescoring, scoring by name/path, and comparing two CLIs.
Quick Start
/printing-press-score # rescore current CLI
/printing-press-score notion-pp-cli-4 # score by name
/printing-press-score ~/my-cli # score by path
/printing-press-score notion-pp-cli-4 vs notion-pp-cli-2 # compare two
Prerequisites
Go 1.26.3 or newer installed
printing-press binary on PATH (install with go install github.com/mvanhorn/cli-printing-press/v4/cmd/printing-press@latest)
Step 0: Setup
Before any other commands, run the setup contract to verify the printing-press binary is on PATH and initialize scope variables:
# min-binary-version: 4.0.0
# Derive scope first — needed for local build detection
_scope_dir="$(git rev-parse --show-toplevel 2>/dev/null || echo "$PWD")"
_scope_dir="$(cd "$_scope_dir" && pwd -P)"
# Prefer local build when running from inside the printing-press repo.
_press_repo=false
if [ -x "$_scope_dir/printing-press" ] && [ -d "$_scope_dir/cmd/printing-press" ]; then
_press_repo=true
export PATH="$_scope_dir:$PATH"
echo "Using local build: $_scope_dir/printing-press"
elif ! command -v printing-press >/dev/null 2>&1; then
if [ -x "$HOME/go/bin/printing-press" ]; then
echo "printing-press found at ~/go/bin/printing-press but not on PATH."
echo "Add GOPATH/bin to your PATH: export PATH=\"\$HOME/go/bin:\$PATH\""
else
echo "printing-press binary not found."
echo "Install with: go install github.com/mvanhorn/cli-printing-press/v4/cmd/printing-press@latest"
fi
return 1 2>/dev/null || exit 1
fi
# Resolve and emit the absolute path the agent must use for every later
# `printing-press` invocation. `export PATH` above only affects this one
# Bash tool call; subsequent calls open a fresh shell and resolve bare
# `printing-press` against the user's default PATH, where a stale global
# can silently shadow the local build. The agent captures this marker and
# substitutes the absolute path into every later invocation.
if [ "$_press_repo" = "true" ]; then
PRINTING_PRESS_BIN="$_scope_dir/printing-press"
else
PRINTING_PRESS_BIN="$(command -v printing-press 2>/dev/null || true)"
fi
echo "PRINTING_PRESS_BIN=$PRINTING_PRESS_BIN"
PRESS_BASE="$(basename "$_scope_dir" | tr '[:upper:]' '[:lower:]' | sed -E 's/[^a-z0-9_-]/-/g; s/^-+//; s/-+$//')"
if [ -z "$PRESS_BASE" ]; then
PRESS_BASE="workspace"
fi
PRESS_SCOPE="$PRESS_BASE-$(printf '%s' "$_scope_dir" | shasum -a 256 | cut -c1-8)"
PRESS_HOME="$HOME/printing-press"
PRESS_RUNSTATE="$PRESS_HOME/.runstate/$PRESS_SCOPE"
PRESS_LIBRARY="$PRESS_HOME/library"
PRESS_MANUSCRIPTS="$PRESS_HOME/manuscripts"
PRESS_CURRENT="$PRESS_RUNSTATE/current"
mkdir -p "$PRESS_RUNSTATE" "$PRESS_LIBRARY" "$PRESS_MANUSCRIPTS" "$PRESS_CURRENT"
After running the setup contract, capture the PRINTING_PRESS_BIN=<abs-path> line from stdout. Every subsequent printing-press ... invocation in this skill must use that absolute path (substitute the value, not the literal $PRINTING_PRESS_BIN token) — export PATH above only affects the single Bash tool call it runs in, so later calls open a fresh shell where bare printing-press resolves against the user's default PATH and a stale global can shadow the local build.
After capturing the binary path, check binary version compatibility. Read the min-binary-version field from this skill's YAML frontmatter. Run <PRINTING_PRESS_BIN> version --json and parse the version from the output. Compare it to min-binary-version using semver rules. If the installed binary is older than the minimum, stop immediately and tell the user: "printing-press binary vX.Y.Z is older than the minimum required vA.B.C. Run go install github.com/mvanhorn/cli-printing-press/v4/cmd/printing-press@latest to update."
Current-run state is resolved from $PRESS_RUNSTATE. Published CLIs are resolved from $PRESS_LIBRARY. Archived manuscripts are resolved from $PRESS_MANUSCRIPTS.
Step 1: Parse Arguments
Read the user's input after /printing-press-score. The input is free-form — interpret intent, don't enforce syntax.
Noise words to strip: compare, vs, versus, and, against, with, to
After stripping noise words, count the remaining tokens:
0 tokens → Rescore Current mode
1 token → Score Single mode
2 tokens → Compare mode
Step 2: Resolve CLI Directories
For each CLI identifier, resolve it to a directory path:
If the token contains / or .
Treat it as a path (absolute or relative). Verify the directory exists.
If the token is a plain name
Try these locations in order:
$PRESS_LIBRARY/<name>/ — exact match
$PRESS_LIBRARY/<name>-pp-cli/ — with -pp-cli suffix
If neither exists, Glob $PRESS_LIBRARY/<name>-pp-cli*
If exactly one glob match exists and is a directory, use it
If multiple glob matches exist, present a numbered menu using AskUserQuestion
If neither exists, scan current-run and archived state:
6. Use Glob to find $PRESS_RUNSTATE/runs/*/state.json files
7. Read each, look for an output_dir or working_dir value whose basename contains the name
8. If found and the directory exists, use it
If nothing resolves, report the error: "Could not find CLI ''. Provide a path or check the name."
Rescore Current (0 tokens)
Use Glob to find all $PRESS_CURRENT/*.json files
Read each to get api_name, state_path, and working_dir
Filter to those whose working_dir actually exists on disk
If none are found, Glob $PRESS_LIBRARY/*-pp-cli* and use those directories instead
If exactly one → use it automatically
If multiple → present a numbered menu using AskUserQuestion:
Multiple CLIs found. Which one to score?
1. stripe-pp-cli ($PRESS_LIBRARY/stripe-pp-cli)
2. notion-pp-cli ($PRESS_LIBRARY/notion-pp-cli)
3. linear-pp-cli ($PRESS_LIBRARY/linear-pp-cli)
If none found → report: "No generated CLIs found. Provide a name or path."
Step 3: Find Spec for Tier 2 Scoring
For each resolved CLI directory, find the OpenAPI spec:
Check <cli-dir>/spec.json — the pipeline converts YAML specs to JSON during generation
If not found, scan $PRESS_RUNSTATE/runs/*/state.json files for one matching this CLI's directory. Read its spec_path field. If that file exists on disk, use it.
If no spec found, proceed without --spec. Note to the user: "No spec found — spec-derived dimensions will be marked N/A and omitted from the denominator. Provide a spec path for full scoring."
Step 4: Run Scorecard
Single Score Mode
Run the scorecard command:
printing-press scorecard --dir <resolved-path> --json
If a spec was found, add --spec <spec-path>.
Parse the JSON output. The structure is:
{
"api_name": "...",
"steinberger": {
"output_modes": 8,
"auth": 7,
"error_handling": 6,
"terminal_ux": 9,
"readme": 5,
"doctor": 10,
"agent_native": 7,
"local_cache": 4,
"breadth": 7,
"vision": 6,
"workflows": 3,
"insight": 5,
"path_validity": 0,
"auth_protocol": 0,
"data_pipeline_integrity": 7,
"sync_correctness": 6,
"type_fidelity": 4,
"dead_code": 3,
"total": 72,
"percentage": 72
},
"overall_grade": "B",
"gap_report": ["..."],
"unscored_dimensions": ["path_validity", "auth_protocol"]
}
If unscored_dimensions is present, those dimensions should be rendered as N/A, not 0/x, and should be described as omitted from the denominator rather than as fixable CLI defects. For backward compatibility, JSON still encodes the numeric fields as 0; consumers must use unscored_dimensions to distinguish N/A from a real zero.
Compare Mode
Run both scorecard commands in parallel using two simultaneous Bash tool calls:
# Call 1:
printing-press scorecard --dir <path1> --spec <spec1> --json
# Call 2:
printing-press scorecard --dir <path2> --spec <spec2> --json
Parse both JSON outputs.
Step 5: Render Output
Single Score Table
Render a rich markdown table. Note: Tier 1 dimensions are all /10. Tier 2 dimensions are /10 except TypeFidelity and DeadCode which are /5.
Scorecard: <api_name>
Infrastructure (Tier 1)
| Dimension | Score |
|----------------|-------|
| Output Modes | 8/10 |
| Auth | 7/10 |
| Error Handling | 6/10 |
| Terminal UX | 9/10 |
| README | 5/10 |
| Doctor | 10/10 |
| Agent Native | 7/10 |
| Local Cache | 4/10 |
| Breadth | 7/10 |
| Vision | 6/10 |
| Workflows | 3/10 |
| Insight | 5/10 |
Domain Correctness (Tier 2)
| Dimension | Score |
|--------------------------|-------|
| Path Validity | 9/10 |
| Auth Protocol | 8/10 |
| Data Pipeline Integrity | 7/10 |
| Sync Correctness | 6/10 |
| Type Fidelity | 4/5 |
| Dead Code | 3/5 |
**Total: 72/100 — Grade B**
If gap_report is non-empty, list the gaps:
Gaps:
- <gap 1>
- <gap 2>
If unscored_dimensions is non-empty, add a note after the table:
Note: path_validity, auth_protocol were unscored and omitted from the denominator. Provide a spec path for full scoring.
Compare Table
Render a side-by-side table with a delta column. Show the first CLI name and second CLI name as column headers. Calculate delta as (CLI 1 score - CLI 2 score). Show +N for positive, -N for negative, — for zero.
Scorecard Comparison: <name1> vs <name2>
Infrastructure (Tier 1)
| Dimension | <name1> | <name2> | Delta |
|----------------|---------|---------|-------|
| Output Modes | 8/10 | 5/10 | +3 |
| Auth | 7/10 | 7/10 | — |
| ... | | | |
Domain Correctness (Tier 2)
| Dimension | <name1> | <name2> | Delta |
|--------------------------|---------|---------|-------|
| Path Validity | 9/10 | 6/10 | +3 |
| ... | | | |
| **Total** | **72/100 (B)** | **56/100 (C)** | **+16** |
Error Handling
If the printing-press binary is not on PATH → show install instructions: go install github.com/mvanhorn/cli-printing-press/v4/cmd/printing-press@latest
If the scorecard command fails → report the error with the full stderr output
If a CLI directory doesn't exist → report which name couldn't be resolved
If JSON parsing fails → show the raw output and report the parsing errordon't have the plugin yet? install it then click "run inline in claude" again.