[PR #702] [MERGED] feat(ghidra): admit explicit DOS COM analysis #2182

New issue

Closed

opened 2026-10-16 21:15:21 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-16 21:15:21 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/702
Author: @N0zoM1z0
Created: 10/6/2026
Status: ✅ Merged
Merged: 10/6/2026
Merged by: @N0zoM1z0

Base: main ← Head: feat/ghidra-dos-com-analysis

📝 Commits (2)

ad9739e feat(ghidra): admit explicit DOS COM analysis
8c5e3b5 fix(verification): validate COM package selection before allocating fixtures

📊 Changes

45 files changed (+1104 additions, -75 deletions)

View changed files

📝 bridge/ghidra/ReaGhidraBridge.java (+23 -0)
➕ bridge/ghidra/ReaGhidraPrepareCom.java (+32 -0)
📝 docs/ghidra-dos.md (+40 -3)
📝 docs/product-catalog.json (+3 -3)
📝 docs/testing.md (+7 -0)
📝 docs/verification/managed-conformance-manifest.json (+2 -2)
📝 package.json (+2 -0)
➕ scripts/verify-real-ghidra-com.mjs (+277 -0)
📝 skills/reverse-engineer-anything/SKILL.md (+2 -2)
📝 skills/reverse-engineer-anything/references/native-and-artifacts.md (+5 -0)
📝 src/application/BinarySessionOpen.ts (+3 -0)
📝 src/application/BinarySessionPort.ts (+2 -0)
📝 src/application/BinaryTargetResolver.ts (+27 -2)
📝 src/application/DirectAnalysis.ts (+20 -1)
📝 src/cli/coreAnalysisCommands.ts (+134 -21)
📝 src/cli/options.ts (+11 -0)
📝 src/contracts/officialToolContracts.ts (+1 -1)
📝 src/contracts/sessionLifecycleInputs.ts (+6 -0)
📝 src/contracts/sessionToolContracts.ts (+1 -1)
📝 src/contracts/toolOutputSchemaPrimitives.ts (+1 -0)

...and 25 more files

📄 Description

Changes

Admit headerless DOS COM only when the caller explicitly selects its interpretation: open_binary({path, format: "dos-com"}) or CLI --target-format dos-com. Omission preserves header detection; the hint intentionally selects a complete-file interpretation independently of its filename. Validate the 64 KiB segment / 256-byte PSP format extent and reject contradictory target kinds.
Import with BinaryLoader at 1000:0100, the x86 real-mode language and default compiler spec. A packaged pre-analysis script seeds the entry and CS/DS/ES/SS context. Failed preparation prevents bridge startup.
Extend independent load-image verification to complete COM bytes, source mappings, entry, register context and absence of fixups; return imposed context separately from unverified PSP, stack and runtime state. Include measured entry context in MZ/other observations and bind the new observation representation in the analysis profile to invalidate incompatible cached snapshots.
Preserve DOS identity through Evidence, snapshots, artifact graphs and CLI/MCP boundaries. Hopper explicitly rejects this format. Package the new Java adapter asset and document the workflow/verification lane.

Verification

114 focused tests across COM format/boundary validation, MZ/COM independent comparisons and negative controls, launcher arguments, provider projection, CLI/catalog contracts and SDK MCP transport.
npm run check:fast; npm run docs:check.
Actual node scripts/verify-real-ghidra-com.mjs: Linux x64 / Ghidra 12.1.4, generated public fixture, CLI/MCP parity for load-image/whole-file bytes/source offsets, imposed entry context, unmapped PSP/partial read, function discovery/decompilation, unchanged original and no surviving owned descendants. Run b2d266c0-bcfb-4efc-874b-a1e57e1285aa.
Actual MZ load-image CLI regression on the source-built MZ fixture, with all checks passing and no surviving owned descendants. Run ec12a784-ea4f-4b57-a424-0acb251e0ba1.
Two private local COM analysis images (original and separately identified decoded derivative) also verified with actual Ghidra. No game data, paths, digests, projects or outputs are committed.
npm package manifest confirms the Java preparation asset is included and private reference/work directories are excluded. The COM lane also passed against an extracted npm tarball with the locked local dependencies, including actual CLI/MCP/decompilation and clean process lineage. Packaged run e975cc01-adb5-43af-8936-ea580bd05fa5.
Real Hopper, macOS and Windows verification was not performed. Windows P0 remains unchanged and excludes DOS.

Examples

rea inspect-native-load-image /absolute/path/to/legacy.com --target-format dos-com --provider ghidra --json
rea function /absolute/path/to/legacy.com 0x10100 --target-format dos-com --provider ghidra --json

{"name":"open_binary","arguments":{"path":"/absolute/path/to/legacy.com","format":"dos-com","provider_id":"ghidra"}}
{"name":"inspect_native_load_image","arguments":{}}

COM is an explicit interpretation, not magic-based format identification. Verified import and decompilation do not establish DOS/PC-98 execution or source reconstruction.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/702 **Author:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Created:** 10/6/2026 **Status:** ✅ Merged **Merged:** 10/6/2026 **Merged by:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Base:** `main` ← **Head:** `feat/ghidra-dos-com-analysis` --- ### 📝 Commits (2) - [`ad9739e`](https://github.com/morluto/rea/commit/ad9739e0e9d8e1e09fb970acca5472bf8f32f7e5) feat(ghidra): admit explicit DOS COM analysis - [`8c5e3b5`](https://github.com/morluto/rea/commit/8c5e3b55442bd7515375a9d440d910cda1179762) fix(verification): validate COM package selection before allocating fixtures ### 📊 Changes **45 files changed** (+1104 additions, -75 deletions) <details> <summary>View changed files</summary> 📝 `bridge/ghidra/ReaGhidraBridge.java` (+23 -0) ➕ `bridge/ghidra/ReaGhidraPrepareCom.java` (+32 -0) 📝 `docs/ghidra-dos.md` (+40 -3) 📝 `docs/product-catalog.json` (+3 -3) 📝 `docs/testing.md` (+7 -0) 📝 `docs/verification/managed-conformance-manifest.json` (+2 -2) 📝 `package.json` (+2 -0) ➕ `scripts/verify-real-ghidra-com.mjs` (+277 -0) 📝 `skills/reverse-engineer-anything/SKILL.md` (+2 -2) 📝 `skills/reverse-engineer-anything/references/native-and-artifacts.md` (+5 -0) 📝 `src/application/BinarySessionOpen.ts` (+3 -0) 📝 `src/application/BinarySessionPort.ts` (+2 -0) 📝 `src/application/BinaryTargetResolver.ts` (+27 -2) 📝 `src/application/DirectAnalysis.ts` (+20 -1) 📝 `src/cli/coreAnalysisCommands.ts` (+134 -21) 📝 `src/cli/options.ts` (+11 -0) 📝 `src/contracts/officialToolContracts.ts` (+1 -1) 📝 `src/contracts/sessionLifecycleInputs.ts` (+6 -0) 📝 `src/contracts/sessionToolContracts.ts` (+1 -1) 📝 `src/contracts/toolOutputSchemaPrimitives.ts` (+1 -0) _...and 25 more files_ </details> ### 📄 Description ## Changes - Admit headerless DOS COM only when the caller explicitly selects its interpretation: `open_binary({path, format: "dos-com"})` or CLI `--target-format dos-com`. Omission preserves header detection; the hint intentionally selects a complete-file interpretation independently of its filename. Validate the 64 KiB segment / 256-byte PSP format extent and reject contradictory target kinds. - Import with BinaryLoader at `1000:0100`, the x86 real-mode language and default compiler spec. A packaged pre-analysis script seeds the entry and CS/DS/ES/SS context. Failed preparation prevents bridge startup. - Extend independent load-image verification to complete COM bytes, source mappings, entry, register context and absence of fixups; return imposed context separately from unverified PSP, stack and runtime state. Include measured entry context in MZ/other observations and bind the new observation representation in the analysis profile to invalidate incompatible cached snapshots. - Preserve DOS identity through Evidence, snapshots, artifact graphs and CLI/MCP boundaries. Hopper explicitly rejects this format. Package the new Java adapter asset and document the workflow/verification lane. ## Verification - 114 focused tests across COM format/boundary validation, MZ/COM independent comparisons and negative controls, launcher arguments, provider projection, CLI/catalog contracts and SDK MCP transport. - `npm run check:fast`; `npm run docs:check`. - Actual `node scripts/verify-real-ghidra-com.mjs`: Linux x64 / Ghidra 12.1.4, generated public fixture, CLI/MCP parity for load-image/whole-file bytes/source offsets, imposed entry context, unmapped PSP/partial read, function discovery/decompilation, unchanged original and no surviving owned descendants. Run `b2d266c0-bcfb-4efc-874b-a1e57e1285aa`. - Actual MZ load-image CLI regression on the source-built MZ fixture, with all checks passing and no surviving owned descendants. Run `ec12a784-ea4f-4b57-a424-0acb251e0ba1`. - Two private local COM analysis images (original and separately identified decoded derivative) also verified with actual Ghidra. No game data, paths, digests, projects or outputs are committed. - npm package manifest confirms the Java preparation asset is included and private reference/work directories are excluded. The COM lane also passed against an extracted npm tarball with the locked local dependencies, including actual CLI/MCP/decompilation and clean process lineage. Packaged run `e975cc01-adb5-43af-8936-ea580bd05fa5`. - Real Hopper, macOS and Windows verification was not performed. Windows P0 remains unchanged and excludes DOS. ## Examples ```bash rea inspect-native-load-image /absolute/path/to/legacy.com --target-format dos-com --provider ghidra --json rea function /absolute/path/to/legacy.com 0x10100 --target-format dos-com --provider ghidra --json ``` ```json {"name":"open_binary","arguments":{"path":"/absolute/path/to/legacy.com","format":"dos-com","provider_id":"ghidra"}} {"name":"inspect_native_load_image","arguments":{}} ``` COM is an explicit interpretation, not magic-based format identification. Verified import and decompilation do not establish DOS/PC-98 execution or source reconstruction. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-16 21:15:21 +00:00

closed this issue
added the
pull-request
label

bsakaguchi referenced this issue

2026-10-31 01:02:27 +00:00

[PR #2182] [MERGED] fix(setup): remove emptied skill directories on uninstall #5070

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

morluto/rea#2182

No description provided.

Rows
Columns