@dotdotgod/opencode 0.0.0-stage → 0.6.7

package/LICENSE

@@ -0,0 +1,94 @@
+Elastic License 2.0 (ELv2)
+
+Copyright 2026 JooYoung Kim
+
+## Acceptance
+
+By using the software, you agree to all of the terms and conditions below.
+
+## Copyright License
+
+The licensor grants you a non-exclusive, royalty-free, worldwide,
+non-sublicensable, non-transferable license to use, copy, distribute, make
+available, and prepare derivative works of the software, in each case subject
+to the limitations and conditions below.
+
+## Limitations
+
+You may not provide the software to third parties as a hosted or managed
+service, where the service provides users with access to any substantial set
+of the features or functionality of the software.
+
+You may not move, change, disable, or circumvent the license key
+functionality in the software, and you may not remove or obscure any
+functionality in the software that is protected by the license key.
+
+You may not alter, remove, or obscure any licensing, copyright, or other
+notices of the licensor in the software. Any use of the licensor's trademarks
+is subject to applicable law.
+
+## Patents
+
+The licensor grants you a license, under any patent claims the licensor can
+license, or becomes able to license, to make, have made, use, sell, offer for
+sale, import and have imported the software, in each case subject to the
+limitations and conditions in this license. This license does not cover any
+patent claims that you cause to be infringed by modifications or additions to
+the software. If you or your company make any written claim that the software
+infringes or contributes to infringement of any patent, your patent license
+for the software granted under these terms ends immediately. If your company
+makes such a claim, your patent license ends immediately for work on behalf
+of your company.
+
+## Notices
+
+You must ensure that anyone who gets a copy of any part of the software from
+you also gets a copy of these terms.
+
+If you modify the software, you must include in any modified copies of the
+software prominent notices stating that you have modified the software.
+
+## No Other Rights
+
+These terms do not imply any licenses other than those expressly granted in
+these terms.
+
+## Termination
+
+If you use the software in violation of these terms, such use is not
+licensed, and your licenses will automatically terminate. If the licensor
+provides you with a notice of your violation, and you cease all violation of
+this license no later than 30 days after you receive that notice, your
+licenses will be reinstated retroactively. However, if you violate these
+terms after such reinstatement, any additional violation of these terms will
+cause your licenses to terminate automatically and permanently.
+
+## No Liability
+
+*As far as the law allows, the software comes as is, without any warranty or
+condition, and the licensor will not be liable to you for any damages arising
+out of these terms or the use or nature of the software, under any kind of
+legal claim.*
+
+## Definitions
+
+The **licensor** is the entity offering these terms, and the **software** is
+the software the licensor makes available under these terms, including any
+portion of it.
+
+**you** refers to the individual or entity agreeing to these terms.
+
+**your company** is any legal entity, sole proprietorship, or other kind of
+organization that you work for, plus all organizations that have control over,
+are under the control of, or are under common control with that organization.
+**control** means ownership of substantially all the assets of an entity, or
+the power to direct its management and policies by vote, contract, or
+otherwise. Control can be direct or indirect.
+
+**your licenses** are all the licenses granted to you for the software under
+these terms.
+
+**use** means anything you do with the software requiring one of your
+licenses.
+
+**trademark** means trademarks, service marks, and similar rights.

package/README.md

@@ -1,3 +1,85 @@
-# Temporary Holding Version
+# OpenCode Adapter
 
-This version is a temporary placeholder for this package. An operational version to replace this has been submitted for review and is awaiting a staged release.
+Dotdotgod project-memory workflows for OpenCode V2: shared MCP execution and
+retrieval, native plan-agent integration, and advisory changed-file impact.
+
+## Start Here
+
+Requires Node.js >=22.5 on PATH. Plan integration targets OpenCode **2.1.2**.
+Other versions retain native plan restrictions and report unavailable integration.
+
+After publishing, add the package to the project's `opencode.jsonc`:
+
+```jsonc
+{
+  "$schema": "https://opencode.ai/config.json",
+  "plugins": ["@dotdotgod/opencode"]
+}
+```
+
+From this source checkout, run `pnpm run generate` first, then use the absolute
+`packages/opencode` directory as the plugin target in a disposable project.
+Local directory loading uses `index.js`; npm loading uses the package exports.
+No project `.opencode/` scaffold or `OPENCODE.md` is required. Keep `AGENTS.md`
+canonical. Reload/restart OpenCode after installation changes.
+
+## Workflows
+
+- `/dd:load`: bounded project map and optional focused query.
+- `/dd:init`: initialize project memory; dry-run by default, writes confirmed.
+- `/dd:impact`: review changed paths through the shared MCP tool.
+- Five shared skills: project-load, doc-first-planning, project-initializer,
+  impact-review, and document-clarify.
+
+There is no `/dd:plan`. Select OpenCode's built-in plan agent. Concrete plans
+are automatically saved to `docs/plan/<task-slug>/README.md` without a separate
+save request, and revised in place as decisions change. Initial questions and
+explanation-only conversations need no file. Saving does not approve execution:
+switch agents before implementation. File creation is performed by the agent's
+normal edit tools, not an automatic transcript dump.
+
+## MCP Tools
+
+The plugin registers `dotdotgod_context` with Code Mode enabled. Discover exact
+tool names from OpenCode's catalog instead of guessing them. It supplies project
+load/initialize/impact, execution, file processing, indexing, search, fetch,
+diagnostics, and ingestion jobs. MCP replies retain content, structuredContent
+and isError; inspect actual execution status and source IDs before claiming success.
+Use auto output for unknown-size commands and bounded source-scoped retrieval.
+Use native read for short, located source and images.
+
+The nineteen shared catalog names have no `dotdotgod_` prefix: for example,
+`project_load`, `project_impact`, `impact_status`, and `execute`. The server namespace
+remains `dotdotgod_context`; no forwarding aliases are registered. Reload sessions
+after upgrading and update saved calls/prompts that use removed names.
+
+## Advisory Impact
+
+Successful observed edit/write/patch operations add project-relative pending
+paths. State persists per host session in `.dotdotgod/context/impact.sqlite`.
+Successful MCP impact clears only covered paths whose revision and content did
+not change during review. Failed reviews retain pending state. The status tool
+reports a bounded list; no pending paths are silently evicted.
+Records expire 24 hours after their revision is stored; expiration is not review.
+
+Pending impact **never blocks** tests, builds, edits, commits or publishing.
+Native permissions and plan-safety restrictions still apply. Shell/script,
+external editor, and unrecognized tool changes are not automatically attributed;
+provide their paths explicitly for review. Existing user dirty files are not
+automatically imported from the whole worktree.
+
+## Compatibility And Verification
+
+Plan redirection recognizes the shipped reminder and replaces the shipped path
+exception in place, preserving later user denies. Unsupported layouts warn rather
+than force-allowing permissions. This is version-coupled integration, not an
+OpenCode public path setting. MCP execution remains write-capable and is not a
+filesystem sandbox; shell permission checks on MCP command strings do not
+reproduce the host's full native shell scanner.
+
+Run `pnpm --filter @dotdotgod/opencode verify`. The opt-in source-checkout command
+`pnpm --filter @dotdotgod/opencode smoke` uses an installed OpenCode binary and a
+local deterministic model endpoint, isolated HOME/XDG directories and its own
+service port; it makes no paid model calls and stops only its isolated service.
+Use `node packages/opencode/scripts/smoke.mjs <directory>` with an extracted
+package to repeat that host smoke against a tarball instead of source files.

package/commands/impact.md

@@ -0,0 +1,26 @@
+---
+description: Review changed-file impact
+---
+
+<!-- Generated from packages/shared resources by scripts/generate-adapters.mjs. Do not edit this file directly; edit the shared source and run `pnpm run generate`. -->
+
+# Review changed-file impact
+
+Use the dotdotgod_context MCP server for project load, initialization, graph impact, execution and indexed retrieval. Discover its exact tool names in OpenCode Code Mode; inspect MCP isError and structuredContent and every execution results entry. For unknown output size use execute with a commands array and outputMode auto; search indexed.id with bounded source-scoped search. Run ancillary CLI checks through MCP execute rather than a duplicate native service. Keep direct read for short located source and images. Pending impact is advisory, never an execution gate. Preserve host permissions. Use the native plan agent, not dd:plan; concrete plans are automatically maintained in docs/plan/<task-slug>/README.md without a separate save request.
+
+User arguments: $ARGUMENTS
+
+## Goal
+
+Review task-relevant source, config, and documentation changes with dotdotgod impact evidence before broad verification, commits, pushes, publishing, or final handoff. OpenCode tracks observed edits per session; pending impact is advisory and never blocks tools.
+
+## Workflow
+
+1. Build the changed-file set from Git status. Preserve unrelated user work and exclude dependencies, caches, build output, secrets, and unrelated local memory.
+2. Call the shared MCP project_impact tool with up to twenty unique paths in first-seen order. Split larger sets into ordered batches of twenty. Inspect isError and structuredContent; only successful reviews of covered unchanged revisions clear pending state. Failures, cancellation and concurrent re-edits retain pending. When MCP is unavailable, follow README indexes, traceability, package metadata and focused search without claiming pending was cleared.
+3. Review the combined ranking and each seed's non-seed top results. Prioritize high-signal specs, tests, architecture, and source links, especially `implemented_by`, `verified_by`, and `related_doc` relationships.
+4. Resolve stale or contradictory related docs and tests that belong to the current task.
+5. Select focused tests, documentation validation, generated-resource checks, and package dry-runs from the evidence. For dotdotgod documentation changes, run `dotdotgod validate . --include-local-memory` and add `--check-index` when index freshness matters.
+6. Report the changed files reviewed, the strongest related findings and follow-up actions, and verification completed or skipped.
+
+Keep raw impact payloads out of normal summaries. Optional hooks may remind agents about this workflow; default package resources leave full verification, index rebuilds, initialization, archive moves, and write blocking under explicit workflow control.

package/commands/init.md

@@ -0,0 +1,38 @@
+---
+description: Initialize project memory
+---
+
+<!-- Generated from packages/shared resources by scripts/generate-adapters.mjs. Do not edit this file directly; edit the shared source and run `pnpm run generate`. -->
+
+# Initialize project memory
+
+Use the dotdotgod_context MCP server for project load, initialization, graph impact, execution and indexed retrieval. Discover its exact tool names in OpenCode Code Mode; inspect MCP isError and structuredContent and every execution results entry. For unknown output size use execute with a commands array and outputMode auto; search indexed.id with bounded source-scoped search. Run ancillary CLI checks through MCP execute rather than a duplicate native service. Keep direct read for short located source and images. Pending impact is advisory, never an execution gate. Preserve host permissions. Use the native plan agent, not dd:plan; concrete plans are automatically maintained in docs/plan/<task-slug>/README.md without a separate save request.
+
+User arguments: $ARGUMENTS
+
+## Goal
+
+Create a non-destructive dotdotgod baseline with canonical agent instructions, thin agent entrypoints, documentation indexes, a project config selected from an appropriate initialization template, and local-memory ignore rules.
+
+## Template Selection
+
+1. Inspect existing agent instructions, the root README, docs indexes, major top-level directories, config, and ignore rules.
+2. Discover custom templates as individual JSON files under `~/.dotdotgod/templates/`. Built-in templates are `software`, `research`, `case-and-evidence`, `publication`, `portfolio`, and `policy`.
+3. Choose from project evidence:
+   - `software`: application, library, or infrastructure projects organized around specs, architecture, source, and tests.
+   - `research`: research diaries, dated records, reports, artifacts, and evaluation evidence.
+   - `case-and-evidence`: canonical case records, evidence or legal grounds, and case outputs.
+   - `publication`: briefs, outlines, chapters, claims, manuscripts, and research sources.
+   - `portfolio`: strategy, positions, ledgers, journals, and market research.
+   - `policy`: policy sections, integrated proposals, evidence, and submission outputs.
+4. Consider custom templates by filename and config contents. A same-name custom template replaces the built-in template completely.
+5. Ask the user when multiple choices are plausible and materially change memory areas or traceability. With insufficient evidence, use `~/.dotdotgod/config.json` `defaultTemplate`, then `software`.
+6. Report the selected template and a concise reason.
+
+## Workflow
+
+1. Preserve existing files and unrelated user work.
+2. Call the shared MCP project_initialize tool with the selected template. Keep dryRun true initially; after explicit write approval call with dryRun false and confirmWrite true. Existing files are skipped, never replaced. Switch out of the native plan agent before initialization.
+3. Use bundled scripts only as an explicit fallback when MCP is unavailable. Custom templates require the CLI; do not silently substitute a bundled template.
+4. Validate the initialized project with `dotdotgod validate <project-root>` when available.
+5. Report created and skipped files, the selected template, validation failures, and unresolved instruction conflicts.

package/commands/load.md

@@ -0,0 +1,28 @@
+---
+description: Load project memory
+---
+
+<!-- Generated from packages/shared resources by scripts/generate-adapters.mjs. Do not edit this file directly; edit the shared source and run `pnpm run generate`. -->
+
+# Load project memory
+
+Use the dotdotgod_context MCP server for project load, initialization, graph impact, execution and indexed retrieval. Discover its exact tool names in OpenCode Code Mode; inspect MCP isError and structuredContent and every execution results entry. For unknown output size use execute with a commands array and outputMode auto; search indexed.id with bounded source-scoped search. Run ancillary CLI checks through MCP execute rather than a duplicate native service. Keep direct read for short located source and images. Pending impact is advisory, never an execution gate. Preserve host permissions. Use the native plan agent, not dd:plan; concrete plans are automatically maintained in docs/plan/<task-slug>/README.md without a separate save request.
+
+User arguments: $ARGUMENTS
+
+## Goal
+
+Load the repository's dotdotgod project memory without modifying maintained files. Treat free-form arguments as query text, not mode switches.
+
+## Workflow
+
+1. Identify the repository root. Expose `Help: dotdotgod --help` as optional CLI discovery guidance without checking or reporting installation status, and do not require Help execution for Load to continue.
+2. When baseline context is missing, read `AGENTS.md`, the adapter entrypoint, `README.md`, and `docs/README.md` when present.
+3. Without arguments, call the shared MCP project_load tool with an empty focus and maxDepth 5. Its configured map owns documentation-root resolution, exclusions and boundary summaries.
+4. With arguments, call project_load with the arguments as focus, limit 30 and maxDepth 3. It returns the configured map plus documentation/code query results and retrieval warnings.
+5. If the CLI or shell execution is unavailable, preserve Load behavior by building the same tree from the configured documentation root when known, excluding local memory, and using README routing. At the depth boundary, list every direct Markdown file and immediate child directory, then summarize hidden descendants separately under each named child with exact recursive directory and Markdown-file counts.
+6. Query searches full documentation content and code outlines/comments, not implementation bodies. Preserve result kind, source ranges and retrieval warnings; verify source freshness before reading a symbol's implementation. Read bodies selectively from query results, explicit paths, and README indexes. Read relevant active plans only when needed, use `docs/archive/README.md` (or the configured archive `README.md` under an alternate documentation root) as the history map, and avoid broad generated, dependency, cache, or secret reads.
+
+## Output
+
+Report the project purpose, key rules, relevant documentation routes, and only the active-plan or next-read hints needed for the request. Keep compact loads to a brief refresh.

package/index.js

@@ -0,0 +1 @@
+export { default } from './src/index.mjs';