Claim list
All 136 first topics laid out: who writes what, where it lands, and how far a run has to go to count. Claiming means posting in the matching board with a title starting with 'Claim:'.
One: pick an unclaimed topic. Two: post in its board titled 'Claim: ', stating how you plan to run it. Three: add the run record or error text to the same thread; once it is enough, it becomes a page here and you are credited in the thread.
Getting started
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Install & launch | Get dsh web running on Windows 11 with 0.2.2-rc.1 and record the tokenized launch URL | Page + thread | No |
| Compare the version numbers installed by npm latest and next on macOS | Page + thread | No | |
| Write a three-step launch checklist marking where each step most often stalls | Page + thread | No | |
| Your first task | Use 'list this folder and count the files' as a first smoke test | Page + thread | No |
| Capture the full terminal trace from prompt to first result | Page + thread | No | |
| CLI and Web UI | Compare what dsh and dsh web produce on the same task | Page + thread | No |
| List the common launch flags and their defaults | Page + thread | No | |
| Config & profiles | Sketch the $DSH_HOME layout and mark which files must be backed up | Page + thread | No |
| Create and switch a profile without disturbing the default | Page + thread | No | |
| Upgrade & rollback paths | Log the config-entry diff across a 0.2.1-rc.5 to 0.2.2-rc.1 upgrade | Ledger + thread | One run |
| Map npm latest and next to the rc and alpha lines | Ledger + thread | One run | |
| Start index | List every node in this group with a one-line purpose | Index page | No |
| Keep a list of structural pages ready to write at launch | Index page | No |
Models, permissions & security
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Model & routing | Configure an OpenAI-compatible gateway and confirm the target model with a minimal tool call | Page + thread | One run |
| Capture the error text and the tracing steps when a model ID is wrong | Page + thread | One run | |
| API keys & credentials | Show keeping a key in an environment variable instead of a plaintext config | Page + thread | One run |
| Write an incident checklist for a key committed by mistake | Page + thread | One run | |
| Permissions & allowlists | Compare a recipe's output with file-write permission on and off | Ledger + thread | One run |
| List permissions declared by common plugins with plain risk notes | Ledger + thread | One run | |
| Workspace isolation & backup | Run a recipe on a copied folder and confirm the original is untouched | Page + thread | One run |
| Drill a full recovery after config gets overwritten | Page + thread | One run | |
| Privacy & data boundaries | Use a redacted sample to see which fields show up in the request body | Page + thread | One run |
| Keep a short list of material that should never be pasted into DSH | Page + thread | One run | |
| Access & security index | List every node here with its required fields | Index page | No |
| Keep an open list of permission and security checks | Index page | No |
Workspace & files
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Reading files & formats | Test scanned PDFs to see whether a missing text layer blocks reading | Page + thread | One run |
| Compare xlsx and csv reading when merged cells are present | Page + thread | One run | |
| Outputs & file changes | Verify a source file's hash is unchanged by a 'make a report, do not edit the ledger' recipe | Page + thread | One run |
| Note docx layout differences between LibreOffice and Word | Page + thread | One run | |
| Batches & deep folders | Trial 10 PDFs then scale to 100 and confirm total equals success plus failures | Page + thread | One run |
| Log timing and missed reads on a very deep folder | Page + thread | One run | |
| Cleanup & undo | Keep a post-run checklist of leftover files | Page + thread | One run |
| Log a full recovery from backup after an accidental deletion | Page + thread | One run | |
| Workspace index | List every node here with its output types | Index page | No |
| Keep the file-handling items still awaiting a real run | Index page | No |
Modes & context
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Working with the modes | Run one task in two modes and log how the turn counts differ | Page + thread | One run |
| Build a mode-selection quick reference with conditions only, no verdicts | Page + thread | One run | |
| PTC vs standard | Run standard and PTC three times each on one task and show token and wall-clock time | Experiment + thread | Full record |
| Add a second comparison: batch repetitive actions vs a small single-file task | Experiment + thread | Full record | |
| Context & compaction | Break down one session's context by source with a context panel | Page + thread | One run |
| Note how the same follow-up answer changes before and after compaction | Page + thread | One run | |
| Prompt structure & placeholders | Rewrite a one-line prompt into a full recipe with input, output and checks | Page + thread | One run |
| Standardize how placeholders and 'read failed' labels are written | Page + thread | One run | |
| Glossary | Write a sourced term entry for 'version gate / engines.dsh', 'PTC' and 'profile' | Glossary | No |
| Build a Chinese-English term map | Glossary | No | |
| Modes & context index | List every node here with its comparison design | Index page | No |
| Keep the open mode-related reproduction tasks | Index page | No |
Task recipes
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Office & admin | Build a searchable Markdown index over 100 PDFs without touching the originals | Page + thread | Full record |
| Turn an Excel ledger into a monthly report with recalculable formulas | Page + thread | Full record | |
| Turn meeting notes into a clean, formatted .docx | Page + thread | Full record | |
| Ops & growth | Watch competitor prices weekly and alert only when something changes | Page + thread | Full record |
| Turn a batch of campaign data into a weekly report flagging anomalies | Page + thread | Full record | |
| Product & specs | Group user feedback into a requirements list keeping the source quotes | Page + thread | Full record |
| Generate a checkable acceptance list from a spec | Page + thread | Full record | |
| Design & content | Rework a long article into copy for several channels without changing the facts | Page + thread | Full record |
| Batch-rename images and generate a mapping table | Page + thread | Full record | |
| Research & study | Turn paper abstracts into a citation-backed reference table | Page + thread | Full record |
| Build a revision outline from lecture notes marking which chapters it covers | Page + thread | Full record | |
| Everyday dev tasks | Generate a change summary and review checklist for a small repo | Page + thread | Full record |
| Turn error logs into a searchable troubleshooting index | Page + thread | Full record | |
| Scheduled & automated | Configure a weekly task and verify it actually fires | Page + thread | Full record |
| Note where the automation entry moved once it became an optional plugin | Page + thread | Full record | |
| Migration & interop | Move old Claude Code / Codex conversations into DSH | Page + thread | Full record |
| List which metadata is lost after migration | Page + thread | Full record | |
| Recipe index | Build a recipe lookup table indexed by role and task type | Index page | No |
| Keep a claim list for template-only, untested recipes | Index page | No |
Experiments & cost data
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Experiment method & controls | Write the site's minimal lab-record standard with no invented numbers | Rules page | No |
| Define a raw-record header for 'three runs each' | Rules page | No | |
| Token & time measurements | Show token and wall-clock time for standard vs PTC | Experiment + thread | Full record |
| Log how cache hits change the readings | Experiment + thread | Full record | |
| Cost & billing checks | Run a known-length task and reconcile the meter with the provider bill | Experiment + thread | Full record |
| Record the real unit-price difference during off-peak hours | Experiment + thread | Full record | |
| Context-plugin comparison | Run five context plugins on one task and show what each costs | Experiment + thread | Full record |
| Note the net saving after each plugin's own context overhead | Experiment + thread | Full record | |
| Upgrade drill | Reproduce the 0.2.1-rc.5 to 0.2.2-rc.1 upgrade and log the plugin-entry count | Experiment + thread | Full record |
| Drill a full rollback and log the recovery time | Experiment + thread | Full record | |
| Lab & cost index | Build an experiment lookup table keyed by environment version | Index page | No |
| Keep the open list of claimed-but-unfinished experiments | Index page | No |
Versions & compatibility
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Version lines & channels | Map the impact of the two breaking changes in 0.2.0-rc.1 | Ledger + thread | No |
| Build a table linking rc/alpha version lines to their channels | Ledger + thread | No | |
| Plugin compatibility ledger | Verify 30 common plugins on 0.2.2-rc.1 with an outcome for each | Ledger + thread | Full record |
| Render the ledger as a static table keyed by host version | Ledger + thread | Full record | |
| The engines.dsh gate | Reproduce a version-gate rejection and log plugin state and the Host log | Ledger + thread | One run |
| Document the two fixes: change host version vs install an older plugin build | Ledger + thread | One run | |
| Backup & rollback checklist | Write a checkable pre-upgrade backup list | Ledger + thread | One run |
| Drill a full recovery after config gets overwritten | Ledger + thread | One run | |
| Compatibility index | List every ledger row with its last-checked date | Index page | No |
| Keep a re-test list for rows marked expired | Index page | No |
Errors & fixes
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Error lookup index | Build an error-text lookup table | Index page | No |
| List the first batch of errors marked 'to reproduce' | Index page | No | |
| Gateway & HTTP 400 errors | Reproduce UNKNOWN_TOOL with a minimal tool call and keep the request body | Page + thread | One run |
| Check how developer-role and tool-name formats differ | Page + thread | One run | |
| Startup & access errors | Reproduce a bare 127.0.0.1:3080 rejection and record the tokenized URL | Page + thread | One run |
| Test the link between a settings 403 and browser extensions | Page + thread | One run | |
| Plugin install & enable errors | Use Host logs to separate install failures from enable failures | Page + thread | One run |
| Log recovery steps when a plugin disappears after a restart | Page + thread | One run | |
| Failure & boundary archive | Turn three failed recipes into a boundary archive | Thread | One run |
| Log how a wrong conclusion was corrected once evidence came in | Thread | One run |
Plugin notes
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Plugin intros & distribution | Write a dsh-context intro using the 12-field structure, marked untested | Ledger + thread | Full record |
| Verify dsh-market's install command and the high-privilege entry it adds | Ledger + thread | Full record | |
| Status & provenance | Write the criteria for the four status labels | Rules page | No |
| Add provenance and a status to each existing plugin page | Rules page | No | |
| Plugin permission matrix | List permission declarations and risks for five common plugins | Ledger + thread | One run |
| Explain why a marketplace plugin hands over plugin-selection power | Ledger + thread | One run | |
| Uninstall & rollback | Drill a full install, disable, remove and restart loop | Page + thread | One run |
| Note config keys left behind after uninstall | Page + thread | One run | |
| Plugin index | List the covered plugins with their verification status | Index page | No |
| Keep a claim list of plugins awaiting verification | Index page | No |
Developers & plugin dev
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Dev setup & scaffolding | Start DSH from source and load a local sample plugin | Page + thread | One run |
| Note how edits to a local plugin get reloaded | Page + thread | One run | |
| Plugin anatomy & manifest | Annotate an open-source plugin's manifest field by field | Glossary | One run |
| Show how an overly narrow engines.dsh range gets gated | Glossary | One run | |
| MCP & tool definitions | Register a minimal custom tool and complete one call | Page + thread | One run |
| Log the error when a gateway rejects a tool name | Page + thread | One run | |
| Release & versioning | Write a CHANGELOG template that states supported host versions | Page + thread | One run |
| Ship two builds of one plugin targeting different 0.2.x lines | Page + thread | One run | |
| Developer index | List every node here with its sample repo | Index page | No |
| Keep the open dev items awaiting a real run | Index page | No |
Community
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| Claim board | Put the first 10 reproduction tasks on the claim board | Thread | No |
| Define how open, claimed and merged states move | Thread | No | |
| Contribution guide & templates | Turn the three templates into copy-ready blank forms | Rules page | No |
| Write a merge rule that editors check facts, not conclusions | Rules page | No | |
| Community rules | Give each of the six rules a good and a bad example | Rules page | No |
| Spell out exactly what 'no official capacity' covers | Rules page | No | |
| Contributors & errata | Build a contributor wall credited by merged entries | Rules page | No |
| Write an errata template: page, error, evidence, suggested fix | Rules page | No |
English Corner
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| English index | Keep a node-by-node map between Chinese nodes and /en/ pages | Index page | No |
| Mark which English pages are translated and which are written natively | Index page | No | |
| Recipes in English | Rewrite the 'index 100 PDFs' recipe in English, keeping its acceptance checklist | Page + thread | One run |
| Mark whether an English recipe shares the Chinese page's verification status | Page + thread | One run | |
| Lab notes in English | Render the standard-vs-PTC raw table in English and confirm the numbers match | Experiment + thread | Full record |
| Link each English lab page to its Chinese data source | Experiment + thread | Full record | |
| English discussion | Write posting guidance and archive rules for the English discussion | Thread | No |
| Collect the first English reader FAQs and map each to a Chinese page | Thread | No |
AI Exchange
| Node | Topic | Lands on | Needs a run |
|---|---|---|---|
| AI Q&A | dsh web won't open after installing 0.2.x on Windows — ask AI for steps | Page + thread | No |
| Have AI compare two rc versions on plugin loading | Page + thread | No |
How claiming works
One: pick an unclaimed topic. Two: post in its board titled 'Claim: <topic>', stating how you plan to run it. Three: add the run record or error text to the same thread; once it is enough, it becomes a page here and you are credited in the thread.