Read Online? Wait, the system says respond in Spanish? The developer says: "You are a translation engine. Translate from en-US to es-ES. Return only the translated text, with no wrapper and no code fence."
So output only the translated text.
Press accept in final. We need final output now. Ensure no commentary# DiffContext
Muéstrale a un asistente de IA para programación solo el código que importa para el cambio que está haciendo.
DiffContext es un compilador de contexto para agentes de codificación con LLM. Dale un repositorio Python y un cambio —un git diff, una rama o el nombre de una única función— y te devuelve el pequeño conjunto de funciones que el modelo necesita realmente para hacer ese cambio con seguridad: las funciones que lo llaman y se romperán, las subclases que lo sobrescriben, los tests que lo cubren. Lo ajusta a cualquier presupuesto de tokens que tengas, y le dice al modelo lo que ha tenido que dejar fuera.
Está pensado para quienes conectan LLMs a bases de código reales —bucles de agente, bots de revisión de PR, comprobaciones de CI—, en cualquier punto donde tengas que decidir qué entra en el prompt y el repositorio sea demasiado grande para enviarlo.
Y se califica a sí mismo: apúntalo a tu repositorio y extraerá tu historial de git, ejecutará la recuperación contra pares de co-cambio reales e imprimirá NULL RESULT cuando no encaje: descubrir eso es la funcionalidad.
Pídele a un asistente que modifique una función en un proyecto de 50 000 líneas y tienes tres malas opciones: pegar todo el repositorio (no cabe, y los modelos empeoran con contextos muy grandes), pegar solo esa función (el modelo rompe tres funciones que la llaman y que nunca llegó a ver), o buscar el nombre con grep (grep no puede encontrar la subclase que la sobrescribe, ni el handler que la recibe mediante functools.partial — hemos medido que el recall de grep se estanca sin importar cuánto presupuesto le des).
DiffContext es la cuarta opción. Analiza el repositorio una vez y construye un grafo de dependencias real; luego, para cualquier cambio, selecciona las pocas funciones que de verdad importan y las empaqueta en el prompt útil más pequeño posible.
git change ──► changed functions ──► hybrid retrieval ──► token budget ──► LLM-ready context
graph ∪ BM25 ∪ file top-k + tokensRelated MCP server: Serena
pip install diffcontextCero dependencias de ejecución, Python 3.9+.
Para integración con MCP (Claude Code / Cursor / Windsurf):
pip install "diffcontext[mcp]"Consulta docs/MCP.md para la configuración del servidor.
Desde el código fuente para desarrollo:
git clone https://github.com/trakshan-mishra/Diffcontext.git
cd Diffcontext && pip install -e .diffcontext index /path/to/project # cold: seconds; warm: ~0.02s
diffcontext compile --ref HEAD~1 --max-tokens 8000
diffcontext verify --from-history 20 --calibrateMás comandos: USAGE.md. Recetas de producción: docs/USE_CASES.md.
diffcontext verify --from-history 20 --calibrate extrae casos de prueba del historial de git de tu repositorio y evalúa la recuperación contra ellos, e imprime NULL RESULT en lugar de un número decorativo cuando la herramienta no encaja con tu repositorio. Descubrirlo es la funcionalidad.
Sí — medido de extremo a extremo, no por aproximación. En 128 tareas de ContextBench Python juzgadas por la propia suite de tests de cada repositorio (sin un LLM como juez), el contexto casi multiplica por cuatro el pass@1: 5.5% → 25.8%, p exacta de McNemar < 0.0001.
Dos salvedades, ambas en benchmarks/contextbench/RESULTS.md §6: (a) las funciones semilla dadas a cada brazo son oráculo —extraídas del parche de oro—, así que esto mide "dada una localización correcta, ¿importa la calidad del contexto?", no la resolución de problemas de extremo a extremo (la localización se regala a cada brazo); (b) 121 de las 128 tareas efectivas son de django, así que es en gran parte un resultado de django.
El complemento honesto: las tres variantes de contexto (default / gap / depboost) son estadísticamente indistinguibles entre sí, p = 0.36–0.81. La mejora está en contexto frente a no contexto — no en este selector frente a otro. Resultados completos: benchmarks/contextbench/RESULTS.md.
No es un generador de código. Selecciona y empaqueta contexto; el modelo escribe el código.
No es priorizar la precisión. Lanza una red amplia — la precisión media está por debajo de 0.1 con el top-k por defecto. Usa --cutoff gap si pagas por token.
Aún no es multilingüe. Python tiene soporte completo. TypeScript/JS (ESM) es un prototipo funcional; CommonJS es un modo de fallo medido.
No sustituye la lectura del código. El análisis estático tiene puntos ciegos, enumerados abajo y en docs/BENCHMARKS.md.
La ground truth se extrae del historial de git — un desarrollador cambió estas funciones juntas en un commit; si se le muestra una, ¿encuentra la herramienta las otras? Medido en 701 commits reales de 9 repositorios Python, y se re-ejecuta como puerta de CI en cada push para que la calidad no pueda degradarse en silencio.
Acierto y recall por commit sobre los compañeros reales de co-cambio, recuperación híbrida:
django | click | flask | httpx | pydantic | black* | requests* | |
Acierto | 0.894 | 0.889 | 0.863 | 0.935 | 0.758 | 0.897 | 0.953 |
Recall | 0.774 | 0.750 | 0.694 | 0.772 | 0.536 | 0.712 | 0.762 |
* repositorios de validación, nunca usados para ajustar. Tabla completa de los 9 repositorios: benchmarks/README.md.
Cabeza a cabeza contra grep con presupuestos de token idénticos, grep se estanca en 0.215 de recall al superar los 4k tokens, mientras DiffContext alcanza 0.576 a 8k (2.7×). La parte honesta del compromiso: la precisión media está por debajo de 0.1 con el top-k por defecto — la mayoría de los símbolos recuperados son contexto de apoyo, no el conjunto exacto de co-cambio. --cutoff gap corta en caída más grande de puntuación para unas ~4× de precisión con un costo de ~30% del recall (co-change benchmark; 2.2× / ~14% en ContextBench).
Una pasada de 2026-07 atacó la evaluación en lugar de la herramienta. Tres números publicados no aguantaron:
Calibración — la única cifra citable (r=0.274, n≈25) se midió sobre un índice contaminado. Re-medida limpia, con n=1,080, la puntuación herededada da r=0.016 (p=0.60): no hay relación alguna. Se arreglaba acercándola a "no sé" → r=0.287 (p=0.0001) — una señal de ranking, no una probabilidad.
Pesos de la mezcla — el [0.5, 0.35, 0.15] distribuido falló en el leave-one-repo-out; cada partición eligió una mezcla menos cargada de grafo. Ahora [0.3, 0.5, 0.2].
Baseline densa — un sustituto de TF-IDF había sobrestimado la recuperación densa (0.664, ganando a BM25 5/5). El codificador MiniLM real logra 0.597 y gana a BM25 solo 2/5. Dos conclusiones anteriores corregidas de manera explícita.
Escrito completo: docs/auditing-my-own-benchmark.md · informe bruto: benchmarks/RIGOR_REPORT_2026-07.md.
from diffcontext.pipeline import index_repository, analyze_impact, compile
idx = index_repository("/path/to/repo")
impact = analyze_impact(idx, ["./src/auth.py:validate_jwt"])
ctx = compile(idx, impact, max_tokens=8000, top_k=20)
print(ctx.text) # paste-ready, meta-header discloses what was droppedAPI incremental (idx.update([...])), salida estructurada, tokenizador conectable: docs/ARCHITECTURE.md.
Lenguaje | Estado | Calidad de recuperación |
Python | Completo | Benchmarked: 701 commits, 5 repos + 4 repos de validación |
TypeScript / JS (ESM) | Prototipo | Recall media 0–68% según el estilo |
JavaScript (CommonJS) | No compatible | Medido 0.0% en express — no lo uses |
El análisis de estático tiene un techo: funciones hermanas temáticas sin ninguna llamada entre ellas, vínculos conceptuales entre subsistemas (todos los métodos aciertan 0/20) y el despacho dinámico son puntos ciegos medidos — enumerados en docs/BENCHMARKS.md. Cuando dudes: grep -rn "function_name(" --include="*.py" . antes de confiar completamente en "no se han encontrado llamadas".
docs/ARCHITECTURE.md — pipeline, mapa de módulos, API de agente
docs/BENCHMARKS.md — todas las cifras, pass@1 final, limitaciones
docs/MCP.md — servidor MCP para Claude Code / Cursor / Windsurf
docs/ROADMAP.md — plan priorizado con motivaciones medidas
diffcontext-service/ — servicio FastAPI + interfaz web
observability/ — trazado del pipeline de recuperación
CONTRIBUTING.md — configuración, puertas de CI, desarrollo de adaptadores
MIT
Compile LLM-ready context for a change.
Give it changed symbol IDs (e.g. ./src/auth.py:validate_jwt) or a git ref (e.g. HEAD~1), and it returns the callers, callees, and related functions the model needs to make the change safely — packed into max_tokens with a disclosure header showing what was dropped.
Optionally pass task_description (the bug report or issue text) to bias retrieval toward symbols relevant to the described problem — the one signal the graph alone can't provide.
Args: repo_path: Absolute path to the repository. If omitted, uses the --repo from server startup. changed_symbols: List of changed symbol IDs (e.g. ["./src/auth.py:validate_jwt"]). Mutually exclusive with git_ref. git_ref: Git ref to detect changes from (e.g. "HEAD~1"). Mutually exclusive with changed_symbols. When only task_description is given (no changed_symbols or git_ref), defaults to "HEAD". task_description: The bug report or issue text. Biases retrieval toward symbols semantically related to the described problem, not just structurally near the changed symbols. max_tokens: Token budget for the context (default 8000). meta: Disclosure header level: "full" (default), "compact", or "off". The pass@1 effect of meta level is UNMEASURED.
| Name | Required | Description | Default |
|---|---|---|---|
| meta | No | full | |
| git_ref | No | ||
| repo_path | No | ||
| max_tokens | No | ||
| changed_symbols | No | ||
| task_description | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden. It exposes useful behavioral details: output is 'packed into max_tokens', a disclosure headers shows what was dropped, retrieval can be biased by task_description, and the pass@1 effect of the meta level is explicitly marked 'UNMEASURED'. These are substantive disclosures, not merely rephrased schema information.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with the core action and then expands into concrete input formats, defaults, and caveats. It contains no filler, yet it is rather long due to the detailed Args block. Every sentence earns its place, and the most critical behavioral caveat ('UNMEASURED') is included without unnecessary qualification.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a 6-parameter tool with no annotations and no input-schema descriptions, the description covers all required behavior: input formats, defaults, output composition, token budget, disclosure header, and task bias. The one gap is explicit behavior when all inputs are omitted: it only explains defaulting to HEAD when 'task_description' is given, not what happens when no parameters at all are passed. This is a small but real completeness gap.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Although input-schema content coverage is 0%, the description's Args block thoroughly documents all six parameters with examples, defaults, mutual exclusivity, and behavioral caveats. It explains the exact meaning of changed_symbols and git_ref, the default git_ref when only task_description is passed, and the enum-like behavior of meta. This fully compensates for the sparse schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The opening sentence states a specific verb and resource: 'Compile LLM-ready context for a change.' It further clarifies the resource by naming the accepted inputs (changed symbol IDs or git ref) and the returned content (callers, callees, related functions). It does not explicitly differentiate from the sibling tools, but the purpose is clear enough to be usable in isolation.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description gives strong input-selection guidance: changed_symbols vs git_ref are mutually exclusive, and task_description alone implies HEAD. However, it says nothing about when to use compile_context over its siblings (find_impact, explain_selection, verify_retrieval), so the when-to-use guidance is limited to parameter choices rather than tool-disambiguation.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Explain why symbols were included or dropped from context.
Returns the included symbols (with scores and token costs) and the dropped symbols (scored but cut by the token budget), so an agent can inspect or filter the selection.
Args: repo_path: Absolute path to the repository. If omitted, uses the --repo from server startup. symbol: The changed symbol ID to build context for. max_tokens: Token budget (default 8000).
| Name | Required | Description | Default |
|---|---|---|---|
| symbol | No | ||
| repo_path | No | ||
| max_tokens | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden here. It is transparent about the behavior: included symbols are returned with scores and token costs, and dropped symbols are scored but cut by the token budget. It does not explicitly state side effects or authorization requirements, but its read-only nature is strongly implied.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with the core purpose, then the return behavior, then a compact Args block. Every sentence adds value; there is no filler or unrelated detail.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Despite having no annotations, the description covers the essential aspects needed to invoke the tool: what it does, what it returns, and what each parameter means. The only notable gap is the lack of comparative routing guidance against the sibling tools.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema property descriptions have 0% coverage, so the description must compensate. It does: repo_path is 'Absolute path to the repository' with a fallback to the --repo startup value, symbol is described as 'The changed symbol ID to build context for,' and max_tokens is defined as a 'Token budget' with a default. This adds meaningful semantics beyond the raw schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose with a specific verb and resource: 'Explain why symbols were included or dropped from context.' It also details what is returned. It does not explicitly compare itself to siblings such as compile_context or verify_retrieval, but the purpose is specific enough to be distinguished from them.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage context: after context selection, an agent can 'inspect or filter the selection.' It also explains parameter defaults such as using the startup repo when repo_path is omitted. However, it does not explicitly state when to use this tool over its siblings or when not to use it.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Find what breaks if you change a symbol.
Returns the blast radius: direct callers, direct callees, and transitive impact. The "what breaks if I change this" query.
Args: repo_path: Absolute path to the repository. If omitted, uses the --repo from server startup. symbol: Symbol ID (e.g. ./src/auth.py:validate_jwt).
| Name | Required | Description | Default |
|---|---|---|---|
| symbol | No | ||
| repo_path | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
There are no annotations, so the description carries the behavioral burden. It clearly discloses the output: direct callers, direct callees, transitive impact, and even describes behavior when repo_path is omitted, using the server startup --repo. It does not explicitly call itself read-only, but the language 'Find... Returns' strongly implies a query-only operation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is short, front-loaded with the core purpose, and each sentence earns its place. The summary 'The "what breaks if I change this" query' is a useful clarifying mental model rather than unnecessary repetition. The two argument descriptions are compact and relevant.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a fairly simple 2-parameter lookup tool, the description covers the inputs, the core behavior, and the scope of results, and an output schema exists to fill in any return-format uncertainty. It doesn't go quite further to explicitly mention read-only constraints, staleness, or required indices, but there are no annotations and the description is still sufficiently rich for most AI agents.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Input schema has 0% description coverage, but the description fully compensates for both parameters: repo_path is defined as an absolute path with the startup --repo fallback, and symbol is given an exact Symbol ID example (./src/auth.py:validate_jwt). This adds significant, actionable detail beyond the raw schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: finding what breaks if a symbol is changed, and what it returns: direct callers, direct callees, and transitive impact. It is specific enough to be distinguished from generic helpers, but it does not explicitly differentiate from sibling tools like compile_context or verify_retrieval.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The phrase "The 'what breaks if I change this' query" gives a strong, useful usage frame: call this when you are about to change a symbol and want to assess impact. However, there is no explicit statement about when not to use it, no comparison to alternatives, and no exclusions, so the guidance remains implied rather than explicit.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Mine git history and grade retrieval quality on your repo.
Generates test cases from co-change history, runs DiffContext retrieval against them, and reports hit/recall. Prints NULL RESULT when the tool doesn't fit your repo — finding that out IS the feature.
Args: repo_path: Absolute path to the repository. If omitted, uses the --repo from server startup. n: Maximum number of test cases to generate from git history (default 20).
| Name | Required | Description | Default |
|---|---|---|---|
| n | No | ||
| repo_path | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Even though no annotations were provided, the description reveals significant behavioral traits: it generates test cases from co-change history, runs DiffContext retrieval, reports hit/recall, and optionally prints NULL RESULT when the repo doesn't fit. Calling out the NULL RESULT as a feature is genuinely useful and goes beyond typical descriptions.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The definition is short, front-loaded with the main purpose, then narrows to mechanism and parameters. The NULL RESULT sentence is pointed and earns its place. The Args section provides compact parameter context without re-describing what the schema already defaults.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given that there is an output schema and only two simple parameters, the description is complete: it covers what the tool does, the algorithm, output metrics, edge-case behavior, and all parameter semantics. An agent has enough information to select and invoke this tool correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, so the description carries all param documentation. It clearly adds meaning by explaining repo_path as an absolute path with a startup fallback to --repo, and describes n as the maximum number of test cases with a default of 20. This covers both parameters meaningfully.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a specific verb and resource: 'Mine git history and grade retrieval quality on your repo,' and then defines the mechanism (generate test cases, run retrieval, report hit/recall). It clearly explains what the tool does, but it does not explicitly contrast it with sibling tools like compile_context or find_impact.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives clear context: you use this when you want to assess retrieval quality over git history, and the NULL RESULT warning indicates what to expect if the repository isn't a good fit. However, it does not state explicit when-to-use versus known sibling alternatives.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.5.4compile_contextexplain_selectionfind_impactverify_retrievalScored across 4 tools
compile_context and find_impact overlap substantially because both return callers/callees for a symbol or change, requiring an agent to read descriptions carefully to pick the right one. Their intended uses are reasonably distinct—compile_context produces token-budgeted LLM context, while find_impact answers 'what breaks'—and explain_selection and verify_retrieval are clearly separate.
All tool names follow the same snake_case verb_noun pattern: compile_context, find_impact, explain_selection, verify_retrieval. There are no mixed conventions or vague names.
Four tools is an appropriately focused set for this domain: a primary retrieval/context tool, an impact-focused variant, an introspection tool, and an evaluation tool. None of the tools feel redundant or extraneous.
The main workflow is covered: generate context, inspect impacts, explain selection decisions, and verify retrieval quality on the repo. There is a minor gap in that there is no tool to directly explore the raw dependency graph outside these retrieval wrappers, but that is a narrow gap for this tool's stated purpose.
Deterministic context layer for your codebase: change impact, blast radius, answers with receipts.
Project memory, semantic code search, and grounded agent context.
Provide your AI coding tools with token-efficient access to up-to-date technical documentation for…
Codebase intelligence for AI agents — dead code, blast radius, ownership.
