decode-minified-js-gates
À propos
Cette compétence analyse du JavaScript minifié pour identifier et classer les implémentations de feature flags, en détectant diverses variantes d'appels de portes et en extrayant leurs valeurs par défaut et leur logique. Elle produit un enregistrement structuré des mécanismes, utile lorsque le comportement des flags n'est pas clair ou lorsque plusieurs bibliothèques de lecture sont utilisées. Les capacités clés incluent l'extraction du contexte, l'identification des variantes, et la détection des conjonctions ou des inversions de kill-switch.
Installation rapide
Claude Code
Recommandénpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/decode-minified-js-gatesCopiez et collez cette commande dans Claude Code pour installer cette compétence
Documentation
Decode Minified JS Gates
Read the call-site context around a flag string in a minified JavaScript bundle and produce a gate-mechanics record: which reader variant, what default, what conjunction, what role. Where probe-feature-flag-state answers "is this gate on or off?", this skill answers the prerequisite question — "what does this gate actually do?"
When to Use
- A flag surfaced by
sweep-flag-namespacecannot be classified from its name alone. - The binary uses more than one gate-reader function and you need to know which one a flag invokes.
- A gate's "default" appears non-boolean (
{},null, a numeric literal) and you need to decode the actual reader variant. - You suspect a kill-switch (inverted gate) but cannot confirm from the flag name.
- A predicate combines multiple gates with
&&and you need to enumerate the co-gates before probing any of them.
Inputs
- Required: a minified JavaScript bundle file (
.js,.mjs,.bun). - Required: a target flag string to decode, in literal form.
- Optional: a list of known reader function names from a prior decode pass — speeds Step 2.
- Optional: a context-window size override; default is 300 chars before, 200 chars after the flag occurrence.
Procedure
Step 1: Extract the Context Window
Locate the flag string and capture an asymmetric window around each occurrence. The pre-context (before the flag) is where the reader function name lives; the post-context (after) is where the default value and conjunction live.
BUNDLE=/path/to/cli/bundle.js
FLAG=acme_widget_v3 # synthetic placeholder
PRE=300
POST=200
# All byte offsets where the flag string occurs
grep -boE "\"${FLAG}\"" "$BUNDLE" | cut -d: -f1 > /tmp/decode-offsets.txt
wc -l /tmp/decode-offsets.txt
# Capture an asymmetric window per occurrence
while read -r offset; do
start=$((offset - PRE))
[ "$start" -lt 0 ] && start=0
length=$((PRE + POST))
echo "=== offset $offset ==="
dd if="$BUNDLE" bs=1 skip="$start" count="$length" 2>/dev/null
echo
done < /tmp/decode-offsets.txt > /tmp/decode-windows.txt
less /tmp/decode-windows.txt
For a fast first pass, grep -oE with negative lookbehind via Perl-compatible regex catches the same windows in one pipe.
Expected: one or more context windows per flag occurrence, each ~500 chars. Multiple occurrences share the same reader function but may differ in default or conjunction — inspect each independently.
On failure: if the bundle is too large for dd-per-occurrence (binary > 100MB or many occurrences), use rg -B 5 -A 3 "$FLAG" "$BUNDLE" for a structured-output approximation. If the windows look corrupted, the bundle may be UTF-16 or have non-ASCII delimiters; use iconv or treat as binary.
Step 2: Identify the Reader Variant
Minified gate libraries commonly expose 4–6 reader variants with different semantics. The reader function name is the first cue; the call signature is the verifier.
The variant taxonomy (synthetic names — substitute the actual minified identifiers from your bundle):
| Variant | Synthetic shape | Returns | Common usage |
|---|---|---|---|
| Sync boolean | gate("flag", false) or gate("flag", true) | boolean | Standard on/off feature switches |
| Sync config-object | fvReader("flag", {key: value}) | JSON object | Structured config (delays, allowlists, model names) |
| Bootstrap-aware TTL | ttlReader("flag", default, ttlMs) | boolean (cached) | Startup-path gates before remote config arrives |
| Truthy-only | truthyReader("flag") | truthy/falsy | Quick checks; no explicit default |
| Async bootstrap | asyncReader("flag") | Promise<boolean> | Gates resolved post-bootstrap |
| Async bridge | bridgeReader("flag") | Promise<boolean> | Bridge/relay-channel gates with separate evaluation path |
Match each context window against the variant patterns:
# Test for variant patterns. Replace the synthetic reader names with the
# actual minified identifiers found in the bundle.
grep -oE '\b(gate|fvReader|ttlReader|truthyReader|asyncReader|bridgeReader)\("acme_widget_v3"' /tmp/decode-windows.txt | sort | uniq -c
If multiple variants appear for the same flag (rare but real — a flag read both sync at startup and async post-bootstrap), record each occurrence's variant separately. Probe results may differ.
Expected: every gate-call occurrence is tagged with one variant. Variant counts across the whole sweep produce a binary-level distribution (e.g., "60% sync boolean, 30% config-object, 10% TTL").
On failure: if a context window contains no recognizable reader pattern, the flag may not be gate-called — recheck the call-site classification from sweep-flag-namespace Step 2. If a window contains a reader name not in this taxonomy, document it as a new variant in your research artifacts and decide whether it warrants a separate handling path.
Step 3: Extract the Default Value
The default is the second positional argument to the reader (or absent for truthy-only / async variants). Capture the exact literal — false, true, null, 0, a string, or a JSON config object.
# Boolean default extraction (sync boolean and TTL variants)
grep -oE '\b(gate|ttlReader)\("acme_widget_v3",\s*(true|false)' /tmp/decode-windows.txt
# Config-object default — match the opening brace and capture until the
# matching brace at the same nesting depth. For minified bundles this is
# safe with a non-greedy match because objects rarely span lines.
grep -oE 'fvReader\("acme_widget_v3",\s*\{[^}]*\}' /tmp/decode-windows.txt
# Numeric default (rare but real for TTL or threshold gates)
grep -oE '\b(gate|ttlReader)\("acme_widget_v3",\s*[0-9]+' /tmp/decode-windows.txt
For config-object defaults, inspect the JSON structure — keys often hint at the gate's purpose (e.g., {maxRetries: 3, timeoutMs: 5000} is a retry-policy config, not a feature toggle).
Expected: an exact literal default per occurrence. Booleans are unambiguous; config-objects need a manual read of the structure.
On failure: if a config-object's matching brace falls outside the context window, increase the post-context size in Step 1. If a default appears to be a variable reference (e.g., gate("flag", x)), the default is computed at runtime — note this as DYNAMIC and probe the actual returned value via probe-feature-flag-state.
Step 4: Detect Conjunctions and Kill Switches
Many gates participate in compound predicates. Conjunctions (&&) and inversions (!) change the gate's effective role.
# Conjunction detection: gate-call followed by `&&` and another gate-call
# within the same predicate window
grep -oE '(gate|fvReader|ttlReader|truthyReader|asyncReader|bridgeReader)\("acme_widget_v3"[^)]*\)\s*&&\s*(gate|fvReader|ttlReader|truthyReader|asyncReader|bridgeReader)\("acme_[a-zA-Z0-9_]+"' /tmp/decode-windows.txt
# Kill-switch detection: leading `!` before the gate-call
grep -oE '!\s*(gate|fvReader|ttlReader|truthyReader|asyncReader|bridgeReader)\("acme_widget_v3"' /tmp/decode-windows.txt
For each detected conjunction, list the co-gate flag names. They are now part of the probe scope — if the target flag's evaluation depends on co-gates, probing the target alone produces incomplete state.
For each detected inversion, mark the flag as a kill switch in the gate-mechanics record. Kill switches flip the meaning of the default: a kill switch with default=false is "feature on by default" (because !false === true), while a normal gate with default=false is "feature off by default."
Expected: a conjunction list (possibly empty) and an inversion flag (boolean) per occurrence.
On failure: if a conjunction includes more than 2 co-gates, the predicate is complex enough that the regex misses the structure. Read the context window manually and document the predicate shape verbatim in the gate-mechanics record.
Step 5: Classify the Gate's Role
Synthesize Steps 2–4 into a role classification. Roles drive different probe strategies and different integration risk.
| Role | Signature | Implication |
|---|---|---|
| Feature switch | sync boolean, no inversion, no conjunction | Standard on/off; probe directly |
| Config provider | sync config-object (fvReader) | Read returned object; default-empty {} ≠ feature off |
| Lifecycle guard | bootstrap-aware TTL or async bootstrap | State depends on bootstrap timing; probe at multiple points |
| Kill switch | inverted gate, default-false | Feature on for users by default; flag flips it OFF |
| Conjunction member | any variant with && co-gate | Cannot evaluate alone; co-gates are part of the probe scope |
| Bridge gate | async bridge variant | Probe must occur over the bridge channel, not the main path |
Expected: every gate-call occurrence has exactly one primary role. Some flags appear in multiple roles across occurrences (e.g., a feature switch in one call site, a conjunction member in another) — record each role independently.
On failure: if a role does not fit the table, the binary is using a gate library not yet documented in this skill. Add a row with synthetic identifiers and contribute the variant back to the skill (or a project-specific extension) for future investigators.
Step 6: Produce the Gate-Mechanics Record
Combine the per-flag findings into a structured record. JSONL is convenient because each flag becomes one line, easy to merge with sweep-flag-namespace inventory.
{"flag":"acme_widget_v3","variant":"sync_boolean","default":false,"role":"feature_switch","conjunctions":[],"inverted":false,"occurrences":3}
{"flag":"acme_retry_policy","variant":"sync_config_object","default":{"maxRetries":3,"timeoutMs":5000},"role":"config_provider","conjunctions":[],"inverted":false,"occurrences":1}
{"flag":"acme_legacy_path","variant":"sync_boolean","default":false,"role":"kill_switch","conjunctions":[],"inverted":true,"occurrences":2}
{"flag":"acme_beta_feature","variant":"sync_boolean","default":false,"role":"conjunction_member","conjunctions":["acme_beta_program_active"],"inverted":false,"occurrences":1}
The gate-mechanics record feeds probe-feature-flag-state Step 2 (gate-vs-event disambiguation): the variant + role + conjunction list determines what observations count as evidence of LIVE / DARK / INDETERMINATE state.
Expected: one JSONL record per flag (or per flag-occurrence if a single flag has multiple distinct mechanics). The record is reproducible — running the procedure again against the same binary produces the same record.
On failure: if records vary across runs, an upstream step is non-deterministic. Most often this is the regex in Step 1 missing or over-matching occurrences. Lock the regexes for the duration of a campaign.
Validation
- Step 1 produces one context window per flag occurrence; windows are ~500 chars
- Step 2 tags each occurrence with exactly one reader variant from the taxonomy
- Step 3 captures the exact default literal (boolean, config-object, or DYNAMIC)
- Step 4 surfaces all conjunctions and kill-switch inversions present in the windows
- Step 5 assigns one role per occurrence, drawn from the role table
- Step 6 produces a JSONL gate-mechanics record that diffs cleanly across re-runs
- All worked examples use synthetic placeholders (
acme_*,gate,fvReader, etc.) — no real flag names, real reader names, or real config-object schemas - The record is consumable by
probe-feature-flag-state(same flag identifiers, compatible field names)
Common Pitfalls
- Reading "default" as "behavior": a gate with
default=trueis on by default in this binary, but server-side overrides may flip it. The default tells you the baseline; the runtime probe (probe-feature-flag-state) tells you the state. - Conflating config-object empty default with feature off:
fvReader("flag", {})returns an empty object as the default — but the flag is on (the gate evaluates to truthy). Treating{}as "off" misclassifies config-providers as feature switches. - Missing kill switches: a leading
!before the gate-call inverts the meaning. Skipping Step 4 produces a record that says "default=false, feature off by default" when the truth is "default=false, feature ON by default because of the inversion." - Probing one half of a conjunction: if
acme_widget_v3 && acme_user_in_cohortis the predicate, probing onlyacme_widget_v3and finding it LIVE does not mean the feature is live — the conjunction may still gate it off via the cohort flag. - Trusting reader names across versions: minified identifiers can change between major versions. The taxonomy in Step 2 is by signature (call shape, return type, default position), not by name. When a binary version changes, re-derive the reader names from a fresh decode pass.
- Window too narrow: a 200/100 split misses config-object defaults that span 300+ chars. Defaults of 300/200 or 400/300 are safer; tighten only if the bundle is huge and the window cost matters.
- Leaking real reader names: minified reader names sometimes look like nonsense (
a,b,Yc1) and feel safe to paste verbatim. They are still findings — substitute synthetic placeholders before publishing the methodology.
Related Skills
probe-feature-flag-state— uses the gate-mechanics record to interpret runtime observationssweep-flag-namespace— produces the candidate flag set this skill decodesmonitor-binary-version-baselines— tracks reader-name changes across binary versions; re-derive Step 2 patterns when baselines flipredact-for-public-disclosure— how to publish gate-decoding methodology without exposing real reader names or schemasconduct-empirical-wire-capture— validates the gate-mechanics record against runtime behavior
Dépôt GitHub
Compétences associées
qmd
Développementqmd est un outil CLI de recherche et d'indexation locale qui permet aux développeurs d'indexer et de rechercher dans des fichiers locaux en utilisant une recherche hybride combinant BM25, des embeddings vectoriels et du reranking. Il prend en charge à la fois une utilisation en ligne de commande et un mode MCP (Model Context Protocol) pour l'intégration avec Claude. L'outil utilise Ollama pour les embeddings et stocke les index localement, ce qui le rend idéal pour rechercher dans de la documentation ou des bases de code directement depuis le terminal.
subagent-driven-development
DéveloppementCette compétence exécute des plans de mise en œuvre en déployant un nouveau sous-agent pour chaque tâche indépendante, avec une revue de code entre les tâches. Elle permet une itération rapide tout en maintenant des contrôles de qualité grâce à ce processus de revue. Utilisez-la lorsque vous travaillez sur des tâches principalement indépendantes au sein d'une même session pour assurer une progression continue avec des vérifications de qualité intégrées.
mcporter
DéveloppementLa compétence mcporter permet aux développeurs de gérer et d'appeler des serveurs Model Context Protocol (MCP) directement depuis Claude. Elle fournit des commandes pour lister les serveurs disponibles, appeler leurs outils avec des arguments, et gérer l'authentification ainsi que le cycle de vie du démon. Utilisez cette compétence pour intégrer et tester les fonctionnalités des serveurs MCP dans votre flux de travail de développement.
adk-deployment-specialist
DéveloppementCette compétence déploie et orchestre des agents Vertex AI ADK en utilisant le protocole A2A, gérant la découverte d'AgentCard, la soumission de tâches, et prenant en charge des outils tels que le bac à sable d'exécution de code et la banque de mémoire. Elle permet de construire des systèmes multi-agents avec des modèles d'orchestration séquentiels, parallèles ou en boucle en Python, Java ou Go. Utilisez-la lorsqu'on vous demande de déployer des agents ADK ou d'orchestrer des flux de travail d'agents sur Google Cloud.
