Our sandbox built YOINK, but its test command failed
Our run on 2026-10-02 reviewed commit a262dda. Installation took 34 seconds and the build step took 8 seconds; both returned exit code 0. The test command returned exit code 1 after 12 seconds. The stored parser summary recorded 200 passed tests and 1 failed test. These summaries do not establish exhaustive unique test coverage.
The retained failure is a demo headline assertion in tests/test_demo.py. The actual text included 1 due now, while the expected text stopped at +35 pts overconfident. The visible mismatch concerns headline text. It neither demonstrates damaged notes nor proves that every vault operation works.
Our harness records command success or failure, exit codes, elapsed time and parsed test counts, alongside retained output tails. The supplied record identifies our sandbox and the reviewed commit. The MrKeyoor testing methodology explains this provenance and its limits; the current methodology page does not establish which procedure revision governed this older run. This correction uses the existing record without a new execution.
The record does not establish a separate interactive terminal trial, an Obsidian interoperability exercise, live JSON-RPC settlement or a production deployment. A successful build command does not establish those outcomes. The commit's README says the demo settles from local files and the test suite never opens a socket.
Dated numerical claims are the reason to try it
YOINK settles dated claims in Markdown notes. Its documented claim fields specify the prediction, settlement date, reading source, numerical comparison and confidence. Settlement writes the outcome, reading, timestamp and source information into the note itself. That keeps a prediction and its recorded result together without moving the notes into a proprietary format.
The YOINK README documents JSON files, dated CSV values, public-chain JSON-RPC readings and manually entered values associated with a URL. An unreadable source leaves a claim open. A manual reading preserves the address and number supplied by the user; it does not independently verify that number.
Settled claims contribute to a Brier score and a comparison between stated confidence and accuracy. The README's forecasting figures belong to a fixed demo, not a person's forecasting record. They do not show that using YOINK improves forecasting. The program records and grades predictions; the README says it has no predictive model.
Obsidian can remain the editor; the ledger remains personal
The README describes Markdown files with frontmatter and wikilinks that Obsidian can open. YOINK can therefore be considered a complement to an Obsidian workflow: its terminal adds capture, settlement and scoring around the same folder. Our sandbox record does not independently verify that interoperability. Try representative notes first, especially if they contain complex metadata, because YOINK implements a frontmatter subset rather than full YAML.
The README says whoever holds the ledger file can rewrite it end to end. Hash chaining can expose an individual edit, but it cannot independently prove when a prediction was made. That is a consequential limit if someone else must rely on the record. For personal review, the ledger can still provide a history whose individual edits are detectable.
The README also distinguishes a captured terminal still from three drawn animations. The animations explain the demo data rather than record application behavior. In particular, the animated graph does not establish a separate terminal pane reachable by a key.
Choose alternatives by the workspace you need
Choose zk for a command-line assistant to a plain-text Zettelkasten or personal wiki. Its README describes templates, search, links and editor integration, and explicitly says zk is not a note editor. It is an alternative for managing linked notes; the supplied evidence does not establish an equivalent prediction-settlement workflow.
Choose SilverBullet for a browser-based Markdown editor programmable with Lua. Its README describes wiki links, a built-in database and queries. That makes it an alternative workspace when browser editing and programmability matter more to your decision. The supplied evidence does not establish equivalent forecasting features or automatic migration from a YOINK vault.
Editorial scores and their limits
Setup ease is 4/5: our sandbox completed installation and the build, while interactive setup remains unverified. Documentation is 5/5 within the supplied material: the README explains claim inputs, storage, refusals and explicit limits. This rates documentation detail, not tested feature coverage. Maturity is 2/5: the package manifest identifies version 0.1.0 as beta, and our recorded test command failed. Community is unassessed because the supplied primary extracts do not establish contributor activity or support responsiveness.
The rubric is 1 for a major demonstrated obstacle, 2 for substantial limitations, 3 for usable with material caveats, 4 for strong evidence with a remaining caveat, and 5 for thorough evidence within the stated assessment scope. A null score means the evidence does not support an assessment. These are editorial judgments, not benchmark results.

